Authors: Ju-Young Kim, Ji-Hong Park, Myeongjun Kim, Gun-Woo Kim
Abstract: Smart farming has emerged as a key technology for advancing modern agriculture through automation and intelligent control. However, systems relying on RGB cameras for perception and robotic manipulators for control, common in smart farming, are vulnerable to photometric perturbations such as hue, illumination, and noise changes, which can cause malfunction under adversarial attacks. To address this issue, we propose an explainable adversarial-robust Vision-Language-Action model based on the OpenVLA-OFT framework. The model integrates an Evidence-3 module that detects photometric perturbations and generates natural language explanations of their causes and effects. Experiments show that the proposed model reduces Current Action L1 loss by 21.7% and Next Actions L1 loss by 18.4% compared to the baseline, demonstrating improved action prediction accuracy and explainability under adversarial conditions.
Authors: D. Shainu Suhas, G. Rahul, K. Muni
Abstract: Monocular 3D lane detection remains challenging due to depth ambiguity, occlusion, and temporal instability across frames. Anchor-based approaches such as Anchor3DLane have demonstrated strong performance by regressing continuous 3D lane curves from multi-camera surround views. However, the baseline model still exhibits (i) sensitivity to regression outliers, (ii) weak supervision of global curve geometry, (iii) difficulty in balancing multiple loss terms, and (iv) limited exploitation of temporal continuity. We propose Temporal-Anchor3DLane, an enhanced 3D lane detection framework that extends Anchor3DLane with three key contributions: (1) a set of multi-task loss improvements, including Balanced L1 regression, Chamfer point-set distance, and uncertainty-based loss weighting, together with focal and Dice components for classification and visibility; (2) a lightweight Temporal LSTM Fusion module that aggregates per-anchor features across frames, replacing a heavier Transformer-style temporal fusion; and (3) ESCOP-style training refinements that couple curve-level supervision with temporal consistency. On OpenLane, Temporal-Anchor3DLane improves F1 by +6.2 and yields smoother temporal trajectories, showing that small architectural and loss refinements significantly enhance 3D lane robustness without extra sensors or scaling.
Authors: Tekleab G. Gebremedhin, Hailom S. Asegede, Bruh W. Tesheme, Tadesse B. Gebremichael, Kalayu G. Redae
Abstract: Agriculture supports over 80% of the population in the Tigray region of Ethiopia, where infrastructural disruptions limit access to expert crop disease diagnosis. We present an offline-first detection system centered on a newly curated indigenous cactus-fig (Opuntia ficus-indica) dataset consisting of 3,587 field images across three core symptom classes. Given deployment constraints in post-conflict edge environments, we benchmark three mobile-efficient architectures: a custom lightweight CNN, EfficientNet-Lite1, and the CNN-Transformer hybrid MobileViT-XS. While the broader system contains independent modules for potato, apple, and corn, this study isolates cactus-fig model performance to evaluate attention sensitivity and inductive bias transfer on indigenous morphology alone. Results establish a clear Pareto trade-off: EfficientNet-Lite1 achieves 90.7% test accuracy, the lightweight CNN reaches 89.5% with the most favorable deployment profile (42 ms inference latency, 4.8 MB model size), and MobileViT-XS delivers 97.3% mean cross-validation accuracy, demonstrating that MHSA-based global reasoning disambiguates pest clusters from two dimensional fungal lesions more reliably than local texture CNN kernels. The ARM compatible models are deployed in a Tigrigna and Amharic localized Flutter application supporting fully offline inference on Cortex-A53 class devices, strengthening inclusivity for food security critical diagnostics.
Authors: Jiahao Jiang, Zhangrui Yang, Xuanhan Wang, Jingkuan Song
Abstract: This extended abstract details our solution for the Global Wheat Full Semantic Segmentation Competition. We developed a systematic self-training framework. This framework combines a two-stage hybrid training strategy with extensive data augmentation. Our core model is SegFormer with a Mix Transformer (MiT-B4) backbone. We employ an iterative teacher-student loop. This loop progressively refines model accuracy. It also maximizes data utilization. Our method achieved competitive performance. This was evident on both the Development and Testing Phase datasets.
Authors: Ranjan Sapkota, Konstantinos I. Roumeliotis, Manoj Karkee, Nikolaos D. Tselikas
Abstract: Deep learning has advanced two fundamentally different paradigms for instance segmentation: specialized models optimized through task-specific fine-tuning and generalist foundation models capable of zero-shot segmentation. This work presents a comprehensive comparison between SAM3 (Segment Anything Model, also called SAMv3) operating in zero-shot mode and three variants of Ultralytics YOLO11 (nano, medium, and large) fine-tuned for instance segmentation. The evaluation is conducted on the MinneApple dataset, a dense benchmark comprising 670 orchard images with 28,179 annotated apple instances, enabling rigorous validation of model behavior under high object density and occlusion. Our analysis shows IoU choices can inflate performance gaps by up to 30%. At the appropriate IoU = 0.15 threshold, YOLO models achieve 68.9%, 72.2%, and 71.9% F1, while SAM3 reaches 59.8% in pure zero-shot mode. However, YOLO exhibits steep degradation 48-50 points across IoU ranges whereas SAM3 drops only 4 points, revealing 12 times superior boundary stability of SAM3. This highlights the strength of SAMv3 in mask precision versus specialization in detection completeness of YOLO11. We provide open-source code, evaluation pipelines, and methodological recommendations, contributing to a deeper understanding of when specialized fine-tuned models or generalist foundation models are preferable for dense instance segmentation tasks. This project repository is available on GitHub as https://github.com/Applied-AI-Research-Lab/Segment-Anything-Model-SAM3-Zero-Shot-Segmentation-Against-Fine-Tuned-YOLO-Detectors
Authors: Mahathir Monjur, Shahriar Nirjon
Abstract: Realistic signal generation and dataset augmentation are essential for advancing mmWave radar applications such as activity recognition and pose estimation, which rely heavily on diverse, and environment-specific signal datasets. However, mmWave signals are inherently complex, sparse, and high-dimensional, making physical simulation computationally expensive. This paper presents mmWeaver, a novel framework that synthesizes realistic, environment-specific complex mmWave signals by modeling them as continuous functions using Implicit Neural Representations (INRs), achieving up to 49-fold compression. mmWeaver incorporates hypernetworks that dynamically generate INR parameters based on environmental context (extracted from RGB-D images) and human motion features (derived from text-to-pose generation via MotionGPT), enabling efficient and adaptive signal synthesis. By conditioning on these semantic and geometric priors, mmWeaver generates diverse I/Q signals at multiple resolutions, preserving phase information critical for downstream tasks such as point cloud estimation and activity classification. Extensive experiments show that mmWeaver achieves a complex SSIM of 0.88 and a PSNR of 35 dB, outperforming existing methods in signal realism while improving activity recognition accuracy by up to 7% and reducing human pose estimation error by up to 15%, all while operating 6-35 times faster than simulation-based approaches.
Authors: Tessa Vu
Abstract: Pedestrian heat exposure is a critical health risk in dense tropical cities, yet standard routing algorithms often ignore micro-scale thermal variation. Hot H\'em is a GeoAI workflow that estimates and operationalizes pedestrian heat exposure in H\^o Ch\'i Minh City (HCMC), Vi\d{e}t Nam, colloquially known as S\`ai G\`on. This spatial data science pipeline combines Google Street View (GSV) imagery, semantic image segmentation, and remote sensing. Two XGBoost models are trained to predict land surface temperature (LST) using a GSV training dataset in selected administrative wards, known as ph\u{o}ng, and are deployed in a patchwork manner across all OSMnx-derived pedestrian network nodes to enable heat-aware routing. This is a model that, when deployed, can provide a foundation for pinpointing where and further understanding why certain city corridors may experience disproportionately higher temperatures at an infrastructural scale.
Authors: Yawar Ali, K. Ramachandra Rao, Ashish Bhaskar, Niladri Chatterjee
Abstract: This paper offers openly available microscopic vehicle trajectory (MVT) datasets collected using unmanned aerial vehicles (UAVs) in heterogeneous, area-based urban traffic conditions. Traditional roadside video collection often fails in dense mixed traffic due to occlusion, limited viewing angles, and irregular vehicle movements. UAV-based recording provides a top-down perspective that reduces these issues and captures rich spatial and temporal dynamics. The datasets described here were extracted using the Data from Sky (DFS) platform and validated against manual counts, space mean speeds, and probe trajectories in earlier work. Each dataset contains time-stamped vehicle positions, speeds, longitudinal and lateral accelerations, and vehicle classifications at a resolution of 30 frames per second. Data were collected at six mid-block locations in the national capital region of India, covering diverse traffic compositions and density levels. Exploratory analyses highlight key behavioural patterns, including lane-keeping preferences, speed distributions, and lateral manoeuvres typical of heterogeneous and area-based traffic settings. These datasets are intended as a resource for the global research community to support simulation modelling, safety assessment, and behavioural studies under area-based traffic conditions. By making these empirical datasets openly available, this work offers researchers a unique opportunity to develop, test, and validate models that more accurately represent complex urban traffic environments.
Authors: Futa Waseda, Shojiro Yamabe, Daiki Shiono, Kento Sasaki, Tsubasa Takahashi
Abstract: Large vision-language models (LVLMs) are vulnerable to typographic attacks, where misleading text within an image overrides visual understanding. Existing evaluation protocols and defenses, largely focused on object recognition, implicitly encourage ignoring text to achieve robustness; however, real-world scenarios often require joint reasoning over both objects and text (e.g., recognizing pedestrians while reading traffic signs). To address this, we introduce a novel task, Read-or-Ignore VQA (RIO-VQA), which formalizes selective text use in visual question answering (VQA): models must decide, from context, when to read text and when to ignore it. For evaluation, we present the Read-or-Ignore Benchmark (RIO-Bench), a standardized dataset and protocol that, for each real image, provides same-scene counterfactuals (read / ignore) by varying only the textual content and question type. Using RIO-Bench, we show that strong LVLMs and existing defenses fail to balance typographic robustness and text-reading capability, highlighting the need for improved approaches. Finally, RIO-Bench enables a novel data-driven defense that learns adaptive selective text use, moving beyond prior non-adaptive, text-ignoring defenses. Overall, this work reveals a fundamental misalignment between the existing evaluation scope and real-world requirements, providing a principled path toward reliable LVLMs. Our Project Page is at https://turingmotors.github.io/rio-vqa/.
Authors: Santosh Patapati
Abstract: We introduce CLARGA, a general-purpose multimodal fusion architecture for multimodal representation learning that works with any number and type of modalities without changing the underlying framework. Given a supervised dataset, CLARGA can be applied to virtually any machine learning task to fuse different multimodal representations for processing by downstream layers. On a sample-by-sample basis, CLARGA learns how modalities should inform one another by building an attention weighted graph over their features and passing messages along this graph with a multi-head Graph Attention Network. Not only does this make CLARGA highly adaptive, as it constructs unique graphs for different samples, it makes for efficient fusion with sub-quadratic complexity as the number of modalities grows. Through a learnable mask, it can also adapt to missing modality inputs. The model is trained with a hybrid objective that combines a supervised task loss with contrastive InfoNCE loss, improving cross-modal consistency and robustness to noisy inputs. We demonstrate CLARGA's effectiveness in diverse multimodal representation learning tasks across 7 datasets spanning finance, human-computer interaction, general multimedia classification, and affective computing. It consistently outperforms baselines, state-of-the-art models, and ablations. Additional experiments also demonstrate its robustness to missing inputs and ability to excel on niche tasks. Overall, CLARGA can be easily plugged into machine learning models for effective and efficient learning of representations across a wide variety of tasks.
Authors: Ming-Zher Poh, Shun Liao, Marco Andreetto, Daniel McDuff, Jonathan Wang, Paolo Di Achille, Jiang Wu, Yun Liu, Lawrence Cai, Eric Teasley, Mark Malhotra, Anupam Pathak, Shwetak Patel
Abstract: Subjective well-being is a cornerstone of individual and societal health, yet its scientific measurement has traditionally relied on self-report methods prone to recall bias and high participant burden. This has left a gap in our understanding of well-being as it is expressed in everyday life. We hypothesized that candid smiles captured during natural smartphone interactions could serve as a scalable, objective behavioral correlate of positive affect. To test this, we analyzed 405,448 video clips passively recorded from 233 consented participants over one week. Using a deep learning model to quantify smile intensity, we identified distinct diurnal and daily patterns. Daily patterns of smile intensity across the week showed strong correlation with national survey data on happiness (r=0.92), and diurnal rhythms documented close correspondence with established results from the day reconstruction method (r=0.80). Higher daily mean smile intensity was significantly associated with more physical activity (Beta coefficient = 0.043, 95% CI [0.001, 0.085]) and greater light exposure (Beta coefficient = 0.038, [0.013, 0.063]), whereas no significant effects were found for smartphone use. These findings suggest that passive smartphone sensing could serve as a powerful, ecologically valid methodology for studying the dynamics of affective behavior and open the door to understanding this behavior at a population scale.
Authors: Noorul Wahab, Nasir Rajpoot
Abstract: Automated generation of diagnostic pathology reports directly from whole slide images (WSIs) is an emerging direction in computational pathology. Translating high-resolution tissue patterns into clinically coherent text remains difficult due to large morphological variability and the complex structure of pathology narratives. We introduce MPath, a lightweight multimodal framework that conditions a pretrained biomedical language model (BioBART) on WSI-derived visual embeddings through a learned visual-prefix prompting mechanism. Instead of end-to-end vision-language pretraining, MPath leverages foundation-model WSI features (CONCH + Titan) and injects them into BioBART via a compact projection module, keeping the language backbone frozen for stability and data efficiency. MPath was developed and evaluated on the RED 2025 Grand Challenge dataset and ranked 4th in Test Phase 2, despite limited submission opportunities. The results highlight the potential of prompt-based multimodal conditioning as a scalable and interpretable strategy for pathology report generation.
Authors: Mozhgan Hadadi, Talukder Z. Jubery, Patrick S. Schnable, Arti Singh, Bedrich Benes, Adarsh Krishnamurthy, Baskar Ganapathysubramanian
Abstract: Accurate 3D plant models are crucial for computational phenotyping and physics-based simulation; however, current approaches face significant limitations. Learning-based reconstruction methods require extensive species-specific training data and lack editability. Procedural modeling offers parametric control but demands specialized expertise in geometric modeling and an in-depth understanding of complex procedural rules, making it inaccessible to domain scientists. We present FloraForge, an LLM-assisted framework that enables domain experts to generate biologically accurate, fully parametric 3D plant models through iterative natural language Plant Refinements (PR), minimizing programming expertise. Our framework leverages LLM-enabled co-design to refine Python scripts that generate parameterized plant geometries as hierarchical B-spline surface representations with botanical constraints with explicit control points and parametric deformation functions. This representation can be easily tessellated into polygonal meshes with arbitrary precision, ensuring compatibility with functional structural plant analysis workflows such as light simulation, computational fluid dynamics, and finite element analysis. We demonstrate the framework on maize, soybean, and mung bean, fitting procedural models to empirical point cloud data through manual refinement of the Plant Descriptor (PD), human-readable files. The pipeline generates dual outputs: triangular meshes for visualization and triangular meshes with additional parametric metadata for quantitative analysis. This approach uniquely combines LLM-assisted template creation, mathematically continuous representations enabling both phenotyping and rendering, and direct parametric control through PD. The framework democratizes sophisticated geometric modeling for plant science while maintaining mathematical rigor.
Authors: Qinghao Meng, Chenming Wu, Liangjun Zhang, Jianbing Shen
Abstract: 3D object detection is essential in autonomous driving, providing vital information about moving objects and obstacles. Detecting objects in distant regions with only a few LiDAR points is still a challenge, and numerous strategies have been developed to address point cloud sparsity through densification.This paper presents a joint completion and detection framework that improves the detection feature in sparse areas while maintaining costs unchanged. Specifically, we propose TransBridge, a novel transformer-based up-sampling block that fuses the features from the detection and completion networks.The detection network can benefit from acquiring implicit completion features derived from the completion network. Additionally, we design the Dynamic-Static Reconstruction (DSRecon) module to produce dense LiDAR data for the completion network, meeting the requirement for dense point cloud ground truth.Furthermore, we employ the transformer mechanism to establish connections between channels and spatial relations, resulting in a high-resolution feature map used for completion purposes.Extensive experiments on the nuScenes and Waymo datasets demonstrate the effectiveness of the proposed framework.The results show that our framework consistently improves end-to-end 3D object detection, with the mean average precision (mAP) ranging from 0.7 to 1.5 across multiple methods, indicating its generalization ability. For the two-stage detection framework, it also boosts the mAP up to 5.78 points.
Authors: Alexander Peysakhovich, William Berman, Joseph Rufo, Felix Wong, Maxwell Z. Wilson
Abstract: Cell painting is a popular technique for creating human-interpretable, high-contrast images of cell morphology. There are two major issues with cell paint: (1) it is labor-intensive and (2) it requires chemical fixation, making the study of cell dynamics impossible. We train a diffusion model (Morphological Observation Neural Enhancement Tool, or MONET) on a large dataset to predict cell paint channels from brightfield images. We show that model quality improves with scale. The model uses a consistency architecture to generate time-lapse videos, despite the impossibility of obtaining cell paint video training data. In addition, we show that this architecture enables a form of in-context learning, allowing the model to partially transfer to out-of-distribution cell lines and imaging protocols. Virtual cell painting is not intended to replace physical cell painting completely, but to act as a complementary tool enabling novel workflows in biological research.
Authors: Cl\'ement Fernandes (SAMOVAR, SOP - SAMOVAR, TSP), Wojciech Pieczynski (SAMOVAR, SOP - SAMOVAR, TSP)
Abstract: Transforming bi-dimensional sets of image pixels into mono-dimensional sequences with a Peano scan (PS) is an established technique enabling the use of hidden Markov chains (HMCs) for unsupervised image segmentation. Related Bayesian segmentation methods can compete with hidden Markov fields (HMFs)-based ones and are much faster. PS has recently been extended to the contextual PS, and some initial experiments have shown the value of the associated HMC model, denoted as HMC-CPS, in image segmentation. Moreover, HMCs have been extended to hidden evidential Markov chains (HEMCs), which are capable of improving HMC-based Bayesian segmentation. In this study, we introduce a new HEMC-CPS model by simultaneously considering contextual PS and evidential HMC. We show its effectiveness for Bayesian maximum posterior mode (MPM) segmentation using synthetic and real images. Segmentation is performed in an unsupervised manner, with parameters being estimated using the stochastic expectation--maximization (SEM) method. The new HEMC-CPS model presents potential for the modeling and segmentation of more complex images, such as three-dimensional or multi-sensor multi-resolution images. Finally, the HMC-CPS and HEMC-CPS models are not limited to image segmentation and could be used for any kind of spatially correlated data.
Authors: Jingmin Zhu, Anqi Zhu, James Bailey, Jun Liu, Hossein Rahmani, Mohammed Bennamoun, Farid Boussaid, Qiuhong Ke
Abstract: Zero-shot skeleton-based action recognition (ZS-SAR) is fundamentally constrained by prevailing approaches that rely on aligning skeleton features with static, class-level semantics. This coarse-grained alignment fails to bridge the domain shift between seen and unseen classes, thereby impeding the effective transfer of fine-grained visual knowledge. To address these limitations, we introduce \textbf{DynaPURLS}, a unified framework that establishes robust, multi-scale visual-semantic correspondences and dynamically refines them at inference time to enhance generalization. Our framework leverages a large language model to generate hierarchical textual descriptions that encompass both global movements and local body-part dynamics. Concurrently, an adaptive partitioning module produces fine-grained visual representations by semantically grouping skeleton joints. To fortify this fine-grained alignment against the train-test domain shift, DynaPURLS incorporates a dynamic refinement module. During inference, this module adapts textual features to the incoming visual stream via a lightweight learnable projection. This refinement process is stabilized by a confidence-aware, class-balanced memory bank, which mitigates error propagation from noisy pseudo-labels. Extensive experiments on three large-scale benchmark datasets, including NTU RGB+D 60/120 and PKU-MMD, demonstrate that DynaPURLS significantly outperforms prior art, setting new state-of-the-art records. The source code is made publicly available at https://github.com/Alchemist0754/DynaPURLS
Authors: Sushmita Nath
Abstract: Predictive maintenance is an important sector in modern industries which improves fault detection and cost reduction processes. By using machine learning algorithms in the whole process, the defects detection process can be implemented smoothly. Semiconductor is a sensitive maintenance field that requires predictability in work. While convolutional neural networks (CNNs) such as VGG-19, Xception and Squeeze-Net have demonstrated solid performance in image classification for semiconductor wafer industry, their effectiveness often declines in scenarios with limited and imbalanced data. This study investigates the use of the Data-Efficient Image Transformer (DeiT) for classifying wafer map defects under data-constrained conditions. Experimental results reveal that the DeiT model achieves highest classification accuracy of 90.83%, outperforming CNN models such as VGG-19(65%), SqueezeNet(82%), Xception(66%) and Hybrid(67%). DeiT also demonstrated superior F1-score (90.78%) and faster training convergence, with enhanced robustness in detecting minority defect classes. These findings highlight the potential of transformer-based models like DeiT in semiconductor wafer defect detection and support predictive maintenance strategies within semiconductor fabrication processes.
Authors: Xianghui Xie, Bowen Wen, Yan Chang, Hesam Rabeti, Jiefeng Li, Ye Yuan, Gerard Pons-Moll, Stan Birchfield
Abstract: Accurate capture of human-object interaction from ubiquitous sensors like RGB cameras is important for applications in human understanding, gaming, and robot learning. However, inferring 4D interactions from a single RGB view is highly challenging due to the unknown object and human information, depth ambiguity, occlusion, and complex motion, which hinder consistent 3D and temporal reconstruction. Previous methods simplify the setup by assuming ground truth object template or constraining to a limited set of object categories. We present CARI4D, the first category-agnostic method that reconstructs spatially and temporarily consistent 4D human-object interaction at metric scale from monocular RGB videos. To this end, we propose a pose hypothesis selection algorithm that robustly integrates the individual predictions from foundation models, jointly refine them through a learned render-and-compare paradigm to ensure spatial, temporal and pixel alignment, and finally reasoning about intricate contacts for further refinement satisfying physical constraints. Experiments show that our method outperforms prior art by 38% on in-distribution dataset and 36% on unseen dataset in terms of reconstruction error. Our model generalizes beyond the training categories and thus can be applied zero-shot to in-the-wild internet videos. Our code and pretrained models will be publicly released.
Authors: Chenrui Fan, Yijun Liang, Shweta Bhardwaj, Kwesi Cobbina, Ming Li, Tianyi Zhou
Abstract: While many vision-language models (VLMs) are developed to answer well-defined, straightforward questions with highly specified targets, as in most benchmarks, they often struggle in practice with complex open-ended tasks, which usually require multiple rounds of exploration and reasoning in the visual space. Such visual thinking paths not only provide step-by-step exploration and verification as an AI detective but also produce better interpretations of the final answers. However, these paths are challenging to evaluate due to the large exploration space of intermediate steps. To bridge the gap, we develop an evaluation suite, ``Visual Reasoning with multi-step EXploration (V-REX)'', which is composed of a benchmark of challenging visual reasoning tasks requiring native multi-step exploration and an evaluation protocol. V-REX covers rich application scenarios across diverse domains. V-REX casts the multi-step exploratory reasoning into a Chain-of-Questions (CoQ) and disentangles VLMs' capability to (1) Planning: breaking down an open-ended task by selecting a chain of exploratory questions; and (2) Following: answering curated CoQ sequentially to collect information for deriving the final answer. By curating finite options of questions and answers per step, V-REX achieves a reliable quantitative and fine-grained analysis of the intermediate steps. By assessing SOTA proprietary and open-sourced VLMs, we reveal consistent scaling trends, significant differences between planning and following abilities, and substantial room for improvement in multi-step exploratory reasoning.
Authors: Antonio Guillen-Perez
Abstract: The development of robust Autonomous Vehicles (AVs) is bottlenecked by the scarcity of "Long-Tail" training data. While fleets collect petabytes of video logs, identifying rare safety-critical events (e.g., erratic jaywalking, construction diversions) remains a manual, cost-prohibitive process. Existing solutions rely on coarse metadata search, which lacks precision, or cloud-based VLMs, which are privacy-invasive and expensive. We introduce Semantic-Drive, a local-first, neuro-symbolic framework for semantic data mining. Our approach decouples perception into two stages: (1) Symbolic Grounding via a real-time open-vocabulary detector (YOLOE) to anchor attention, and (2) Cognitive Analysis via a Reasoning VLM that performs forensic scene analysis. To mitigate hallucination, we implement a "System 2" inference-time alignment strategy, utilizing a multi-model "Judge-Scout" consensus mechanism. Benchmarked on the nuScenes dataset against the Waymo Open Dataset (WOD-E2E) taxonomy, Semantic-Drive achieves a Recall of 0.966 (vs. 0.475 for CLIP) and reduces Risk Assessment Error by 40\% compared to single models. The system runs entirely on consumer hardware (NVIDIA RTX 3090), offering a privacy-preserving alternative to the cloud.
Authors: Senhao Gao, Junqing Zhang, Luoyu Mei, Shuai Wang, Xuyu Wang
Abstract: Human activity recognition (HAR) requires extracting accurate spatial-temporal features with human movements. A mmWave radar point cloud-based HAR system suffers from sparsity and variable-size problems due to the physical features of the mmWave signal. Existing works usually borrow the preprocessing algorithms for the vision-based systems with dense point clouds, which may not be optimal for mmWave radar systems. In this work, we proposed a graph representation with a discrete dynamic graph neural network (DDGNN) to explore the spatial-temporal representation of human movement-related features. Specifically, we designed a star graph to describe the high-dimensional relative relationship between a manually added static center point and the dynamic mmWave radar points in the same and consecutive frames. We then adopted DDGNN to learn the features residing in the star graph with variable sizes. Experimental results demonstrated that our approach outperformed other baseline methods using real-world HAR datasets. Our system achieved an overall classification accuracy of 94.27\%, which gets the near-optimal performance with a vision-based skeleton data accuracy of 97.25\%. We also conducted an inference test on Raspberry Pi~4 to demonstrate its effectiveness on resource-constraint platforms. \sh{ We provided a comprehensive ablation study for variable DDGNN structures to validate our model design. Our system also outperformed three recent radar-specific methods without requiring resampling or frame aggregators.
Authors: Tran-Vu La, Minh-Tan Pham, Yu Li, Patrick Matgen, Marco Chini
Abstract: We investigate the application of Federated Learning (FL) for ship detection across diverse satellite datasets, offering a privacy-preserving solution that eliminates the need for data sharing or centralized collection. This approach is particularly advantageous for handling commercial satellite imagery or sensitive ship annotations. Four FL models including FedAvg, FedProx, FedOpt, and FedMedian, are evaluated and compared to a local training baseline, where the YOLOv8 ship detection model is independently trained on each dataset without sharing learned parameters. The results reveal that FL models substantially improve detection accuracy over training on smaller local datasets and achieve performance levels close to global training that uses all datasets during the training. Furthermore, the study underscores the importance of selecting appropriate FL configurations, such as the number of communication rounds and local training epochs, to optimize detection precision while maintaining computational efficiency.
Authors: Maria Rodriguez, Minh-Tan Pham, Martin Sudmanns, Quentin Poterek, Oscar Narvaez
Abstract: After a wildfire, delineating burned areas (BAs) is crucial for quantifying damages and supporting ecosystem recovery. Current BA mapping approaches rely on computer vision models trained on post-event remote sensing imagery, but often overlook their applicability to time-constrained emergency management scenarios. This study introduces a supervised semantic segmentation workflow aimed at boosting both the performance and efficiency of BA delineation. It targets SPOT-6/7 imagery due to its very high resolution and on-demand availability. Experiments are evaluated based on Dice score, Intersection over Union, and inference time. The results show that U-Net and SegFormer models perform similarly with limited training data. However, SegFormer requires more resources, challenging its practical use in emergencies. Incorporating land cover data as an auxiliary task enhances model robustness without increasing inference time. Lastly, Test-Time Augmentation improves BA delineation performance but raises inference time, which can be mitigated with optimization methods like Mixed Precision.
Authors: Tejas Panambur, Ishan Rajendrakumar Dave, Chongjian Ge, Ersin Yumer, Xue Bai
Abstract: Modern text-to-video (T2V) diffusion models can synthesize visually compelling clips, yet they remain brittle at fine-scale structure: even state-of-the-art generators often produce distorted faces and hands, warped backgrounds, and temporally inconsistent motion. Such severe structural artifacts also appear in very low-quality real-world videos. Classical video restoration and super-resolution (VR/VSR) methods, in contrast, are tuned for synthetic degradations such as blur and downsampling and tend to stabilize these artifacts rather than repair them, while diffusion-prior restorers are usually trained on photometric noise and offer little control over the trade-off between perceptual quality and fidelity. We introduce CreativeVR, a diffusion-prior-guided video restoration framework for AI-generated (AIGC) and real videos with severe structural and temporal artifacts. Our deep-adapter-based method exposes a single precision knob that controls how strongly the model follows the input, smoothly trading off between precise restoration on standard degradations and stronger structure- and motion-corrective behavior on challenging content. Our key novelty is a temporally coherent degradation module used during training, which applies carefully designed transformations that produce realistic structural failures. To evaluate AIGC-artifact restoration, we propose the AIGC54 benchmark with FIQA, semantic and perceptual metrics, and multi-aspect scoring. CreativeVR achieves state-of-the-art results on videos with severe artifacts and performs competitively on standard video restoration benchmarks, while running at practical throughput (about 13 FPS at 720p on a single 80-GB A100). Project page: https://daveishan.github.io/creativevr-webpage/.
Authors: Ryan Po, Eric Ryan Chan, Changan Chen, Gordon Wetzstein
Abstract: Autoregressive video models are promising for world modeling via next-frame prediction, but they suffer from exposure bias: a mismatch between training on clean contexts and inference on self-generated frames, causing errors to compound and quality to drift over time. We introduce Backwards Aggregation (BAgger), a self-supervised scheme that constructs corrective trajectories from the model's own rollouts, teaching it to recover from its mistakes. Unlike prior approaches that rely on few-step distillation and distribution-matching losses, which can hurt quality and diversity, BAgger trains with standard score or flow matching objectives, avoiding large teachers and long-chain backpropagation through time. We instantiate BAgger on causal diffusion transformers and evaluate on text-to-video, video extension, and multi-prompt generation, observing more stable long-horizon motion and better visual consistency with reduced drift.
Authors: Guanfang Dong, Luke Schultz, Negar Hassanpour, Chao Gao
Abstract: The superior representation capability of pre-trained vision foundation models (VFMs) has been harnessed for enhancing latent diffusion models (LDMs). These approaches inject the rich semantics from high-dimensional VFM representations (e.g., DINOv3) into LDMs at different phases, resulting in accelerated learning and better generation performance. However, the high-dimensionality of VFM representations may also lead to Information Overload, particularly when the VFM features exceed the size of the original image for decoding. To address this issue while preserving the utility of VFM features, we propose RePack (Representation Packing), a simple yet effective framework for improving Diffusion Transformers (DiTs). RePack transforms the VFM representation into a more compact, decoder-friendly representation by projecting onto low-dimensional manifolds. We find that RePack can effectively filter out non-semantic noise while preserving the core structural information needed for high-fidelity reconstruction. Experimental results show that RePack significantly accelerates DiT convergence and outperforms recent methods that directly inject raw VFM features into the decoder for image reconstruction. On DiT-XL/2, RePack achieves an FID of 3.66 in only 64 epochs, which is 35% faster than the state-of-the-art method. This demonstrates that RePack successfully extracts the core semantics of VFM representations while bypassing their high-dimensionality side effects.
Authors: Zihu Wang, Boxun Xu, Yuxuan Xia, Peng Li
Abstract: Large vision-language models (LVLMs) exhibit impressive ability to jointly reason over visual and textual inputs. However, they often produce outputs that are linguistically fluent but factually inconsistent with the visual evidence, i.e., they hallucinate. Despite growing efforts to mitigate such hallucinations, a key question remains: what form of visual attention can effectively suppress hallucinations during decoding? In this work, we provide a simple answer: the vision encoder's own attention map. We show that LVLMs tend to hallucinate when their final visual-attention maps fail to concentrate on key image objects, whereas the vision encoder's more concentrated attention maps substantially reduce hallucinations. To further investigate the cause, we analyze vision-text conflicts during decoding and find that these conflicts peak in the language model's middle layers. Injecting the vision encoder's attention maps into these layers effectively suppresses hallucinations. Building on these insights, we introduce VEGAS, a simple yet effective inference-time method that integrates the vision encoder's attention maps into the language model's mid-layers and adaptively steers tokens which fail to concentrate on key image objects. Extensive experiments across multiple benchmarks demonstrate that VEGAS consistently achieves state-of-the-art performance in reducing hallucinations.
Authors: Samar Fares, Nurbek Tastan, Karthik Nandakumar
Abstract: The advent of high-quality video generation models has amplified the need for robust watermarking schemes that can be used to reliably detect and track the provenance of generated videos. Existing video watermarking methods based on both post-hoc and in-generation approaches fail to simultaneously achieve imperceptibility, robustness, and computational efficiency. This work introduces a novel framework for in-generation video watermarking called SPDMark (pronounced `SpeedMark') based on selective parameter displacement of a video diffusion model. Watermarks are embedded into the generated videos by modifying a subset of parameters in the generative model. To make the problem tractable, the displacement is modeled as an additive composition of layer-wise basis shifts, where the final composition is indexed by the watermarking key. For parameter efficiency, this work specifically leverages low-rank adaptation (LoRA) to implement the basis shifts. During the training phase, the basis shifts and the watermark extractor are jointly learned by minimizing a combination of message recovery, perceptual similarity, and temporal consistency losses. To detect and localize temporal modifications in the watermarked videos, we use a cryptographic hashing function to derive frame-specific watermark messages from the given base watermarking key. During watermark extraction, maximum bipartite matching is applied to recover the correct frame order, even from temporally tampered videos. Evaluations on both text-to-video and image-to-video generation models demonstrate the ability of SPDMark to generate imperceptible watermarks that can be recovered with high accuracy and also establish its robustness against a variety of common video modifications.
Authors: Swarn S. Warshaneyan, Maksims Ivanovs, Bla\v{z} Cugmas, Inese B\=erzi\c{n}a, Laura Goldberga, Mindaugas Tamosiunas, Roberts Kadi\c{k}is
Abstract: We present a comprehensive study on fully automated pollen recognition across both conventional optical and digital in-line holographic microscopy (DIHM) images of sample slides. Visually recognizing pollen in unreconstructed holographic images remains challenging due to speckle noise, twin-image artifacts and substantial divergence from bright-field appearances. We establish the performance baseline by training YOLOv8s for object detection and MobileNetV3L for classification on a dual-modality dataset of automatically annotated optical and affinely aligned DIHM images. On optical data, detection mAP50 reaches 91.3% and classification accuracy reaches 97%, whereas on DIHM data, we achieve only 8.15% for detection mAP50 and 50% for classification accuracy. Expanding the bounding boxes of pollens in DIHM images over those acquired in aligned optical images achieves 13.3% for detection mAP50 and 54% for classification accuracy. To improve object detection in DIHM images, we employ a Wasserstein GAN with spectral normalization (WGAN-SN) to create synthetic DIHM images, yielding an FID score of 58.246. Mixing real-world and synthetic data at the 1.0 : 1.5 ratio for DIHM images improves object detection up to 15.4%. These results demonstrate that GAN-based augmentation can reduce the performance divide, bringing fully automated DIHM workflows for veterinary imaging a small but important step closer to practice.
Authors: Yuheng Li, Yue Zhang, Abdoul Aziz Amadou, Yuxiang Lai, Jike Zhong, Tiziano Passerini, Dorin Comaniciu, Puneet Sharma
Abstract: Echocardiography is the most widely used imaging modality in cardiology, yet its interpretation remains labor-intensive and inherently multimodal, requiring view recognition, quantitative measurements, qualitative assessments, and guideline-based reasoning. While recent vision-language models (VLMs) have achieved broad success in natural images and certain medical domains, their potential in echocardiography has been limited by the lack of large-scale, clinically grounded image-text datasets and the absence of measurement-based reasoning central to echo interpretation. We introduce EchoGround-MIMIC, the first measurement-grounded multimodal echocardiography dataset, comprising 19,065 image-text pairs from 1,572 patients with standardized views, structured measurements, measurement-grounded captions, and guideline-derived disease labels. Building on this resource, we propose EchoVLM, a vision-language model that incorporates two novel pretraining objectives: (i) a view-informed contrastive loss that encodes the view-dependent structure of echocardiographic imaging, and (ii) a negation-aware contrastive loss that distinguishes clinically critical negative from positive findings. Across five types of clinical applications with 36 tasks spanning multimodal disease classification, image-text retrieval, view classification, chamber segmentation, and landmark detection, EchoVLM achieves state-of-the-art performance (86.5% AUC in zero-shot disease classification and 95.1% accuracy in view classification). We demonstrate that clinically grounded multimodal pretraining yields transferable visual representations and establish EchoVLM as a foundation model for end-to-end echocardiography interpretation. We will release EchoGround-MIMIC and the data curation code, enabling reproducibility and further research in multimodal echocardiography interpretation.
Authors: Dashti A. Ali, Aras T. Asaad, Jacob J. Peoples, Mohammad Hamghalam, Alex Robins, Mane Piliposyan, Richard K. G. Do, Natalie Gangai, Yun S. Chun, Ahmad Bashir Barekzai, Jayasree Chakraborty, Hala Khasawneh, Camila Vilela, Natally Horvat, Jo\~ao Miranda, Alice C. Wei, Amber L. Simpson
Abstract: The development of machine learning (ML) models based on computed tomography (CT) imaging modality has been a major focus of recent research in the medical imaging domain. Incorporating robust feature engineering approach can highly improve the performance of these models. Topological data analysis (TDA), a recent development based on the mathematical field of algebraic topology, mainly focuses on the data from a topological perspective, extracting deeper insight and higher dimensional structures from the data. Persistent homology (PH), a fundamental tool in the area of TDA, can extract topological features such as connected components, cycles and voids from the data. A popular approach to construct PH from 3D CT images is to utilize the 3D cubical complex filtration, a method adapted for grid-structured data. However, this approach may not always yield the best performance and can suffer from computational complexity with higher resolution CT images. This study introduces a novel patch-based PH construction approach tailored for volumetric medical imaging data, in particular CT modality. A wide range of experiments has been conducted on several datasets of 3D CT images to comprehensively analyze the performance of the proposed method with various parameters and benchmark it against the 3D cubical complex algorithm. Our results highlight the dominance of the patch-based TDA approach in terms of both classification performance and time-efficiency. The proposed approach outperformed the cubical complex method, achieving average improvement of 10.38%, 6.94%, 2.06%, 11.58%, and 8.51% in accuracy, AUC, sensitivity, specificity, and F1 score, respectively, across all datasets. Finally, we provide a convenient python package, Patch-TDA, to facilitate the utilization of the proposed approach.
Authors: Thomas Manzini, Priyankari Perali, Raisa Karnik, Robin R. Murphy
Abstract: This paper presents the largest known benchmark dataset for road damage assessment and road alignment, and provides 18 baseline models trained on the CRASAR-U-DRIODs dataset's post-disaster small uncrewed aerial systems (sUAS) imagery from 10 federally declared disasters, addressing three challenges within prior post-disaster road damage assessment datasets. While prior disaster road damage assessment datasets exist, there is no current state of practice, as prior public datasets have either been small-scale or reliant on low-resolution imagery insufficient for detecting phenomena of interest to emergency managers. Further, while machine learning (ML) systems have been developed for this task previously, none are known to have been operationally validated. These limitations are overcome in this work through the labeling of 657.25km of roads according to a 10-class labeling schema, followed by training and deploying ML models during the operational response to Hurricanes Debby and Helene in 2024. Motivated by observed road line misalignment in practice, 9,184 road line adjustments were provided for spatial alignment of a priori road lines, as it was found that when the 18 baseline models are deployed against real-world misaligned road lines, model performance degraded on average by 5.596\% Macro IoU. If spatial alignment is not considered, approximately 8\% (11km) of adverse conditions on road lines will be labeled incorrectly, with approximately 9\% (59km) of road lines misaligned off the actual road. These dynamics are gaps that should be addressed by the ML, CV, and robotics communities to enable more effective and informed decision-making during disasters.
Authors: Bj\"orn L\"utjens, Patrick Alexander, Raf Antwerpen, Til Widmann, Guido Cervone, Marco Tedesco
Abstract: The Greenland ice sheet is melting at an accelerated rate due to processes that are not fully understood and hard to measure. The distribution of surface meltwater can help understand these processes and is observable through remote sensing, but current maps of meltwater face a trade-off: They are either high-resolution in time or space, but not both. We develop a deep learning model that creates gridded surface meltwater maps at daily 100m resolution by fusing data streams from remote sensing observations and physics-based models. In particular, we spatiotemporally downscale regional climate model (RCM) outputs using synthetic aperture radar (SAR), passive microwave (PMW), and a digital elevation model (DEM) over the Helheim Glacier in Eastern Greenland from 2017-2023. Using SAR-derived meltwater as "ground truth", we show that a deep learning-based method that fuses all data streams is over 10 percentage points more accurate over our study area than existing non deep learning-based approaches that only rely on a regional climate model (83% vs. 95% Acc.) or passive microwave observations (72% vs. 95% Acc.). Alternatively, creating a gridded product through a running window calculation with SAR data underestimates extreme melt events, but also achieves notable accuracy (90%) and does not rely on deep learning. We evaluate standard deep learning methods (UNet and DeepLabv3+), and publish our spatiotemporally aligned dataset as a benchmark, MeltwaterBench, for intercomparisons with more complex data-driven downscaling methods. The code and data are available at $\href{https://github.com/blutjens/hrmelt}{github.com/blutjens/hrmelt}$.
Authors: Ayush Vaibhav Bhatti, Deniz Karakay, Debottama Das, Nilotpal Rajbongshi, Yuito Sugimoto
Abstract: Open-world deployment requires models to recognize both known categories and remain reliable when novel classes appear. We present a unified experimental study spanning open-set recognition (OSR) and few-shot class-incremental learning (FSCIL) on CIFAR-10. For OSR, we compare three pretrained frozen visual encoders: ResNet-50, ConvNeXt-Tiny and CLIP ViT-B/16,using a linear probe and four post-hoc scoring functions, namely MSP, Energy, Mahalanobis and kNN. Across metrics,such as, AUROC, AUPR, FPR@95, and OSCR, CLIP consistently yields the strongest separability between known and unknown samples, with Energy providing the most stable performance across backbones. For FSCIL, we compare modified SPPR, OrCo, and ConCM using partially frozen ResNet-50 across 1-, 5-, and 10-shot scenarios. ConCM achieves 84.7% accuracy in the 10-shot setting with the cleanest confusion matrix, while all methods show saturation beyond 5 shots. Our controlled evaluation reveals how the backbone architecture and scoring mechanisms affect unknown detection and how prototype-based methods mitigate catastrophic forgetting during incremental adaptation.
Authors: Daniel Adebi, Sagnik Majumder, Kristen Grauman
Abstract: Understanding camera motion is a fundamental problem in embodied perception and 3D scene understanding. While visual methods have advanced rapidly, they often struggle under visually degraded conditions such as motion blur or occlusions. In this work, we show that passive scene sounds provide complementary cues for relative camera pose estimation for in-the-wild videos. We introduce a simple but effective audio-visual framework that integrates direction-ofarrival (DOA) spectra and binauralized embeddings into a state-of-the-art vision-only pose estimation model. Our results on two large datasets show consistent gains over strong visual baselines, plus robustness when the visual information is corrupted. To our knowledge, this represents the first work to successfully leverage audio for relative camera pose estimation in real-world videos, and it establishes incidental, everyday audio as an unexpected but promising signal for a classic spatial challenge. Project: http://vision.cs.utexas.edu/projects/av_camera_pose.
Authors: Xuancheng Xu, Yaning Li, Sisi You, Bing-Kun Bao
Abstract: Customized video generation aims to produce videos that faithfully preserve the subject's appearance from reference images while maintaining temporally consistent motion from reference videos. Existing methods struggle to ensure both subject appearance similarity and motion pattern consistency due to the lack of object-level guidance for subject and motion. To address this, we propose SMRABooth, which leverages the self-supervised encoder and optical flow encoder to provide object-level subject and motion representations. These representations are aligned with the model during the LoRA fine-tuning process. Our approach is structured in three core stages: (1) We exploit subject representations via a self-supervised encoder to guide subject alignment, enabling the model to capture overall structure of subject and enhance high-level semantic consistency. (2) We utilize motion representations from an optical flow encoder to capture structurally coherent and object-level motion trajectories independent of appearance. (3) We propose a subject-motion association decoupling strategy that applies sparse LoRAs injection across both locations and timing, effectively reducing interference between subject and motion LoRAs. Extensive experiments show that SMRABooth excels in subject and motion customization, maintaining consistent subject appearance and motion patterns, proving its effectiveness in controllable text-to-video generation.
Authors: Ercan Erkalkan, Vedat Topuz, Ay\c{c}a Ak
Abstract: This study introduces a lightweight perimeter tracking method designed for micro UAV teams operating over wildfire environments under limited bandwidth conditions. Thermal image frames generate coarse hot region masks through adaptive thresholding and morphological refinement, while RGB frames contribute edge cues and suppress texture related false detections using gradient based filtering. A rule level merging strategy selects boundary candidates and simplifies them via the Ramer Douglas Peucker algorithm. The system incorporates periodic beacons and an inertial feedback loop that maintains trajectory stability in the presence of GPS degradation. The guidance loop targets sub 50 ms latency on embedded System on Chip (SoC) platforms by constraining per frame pixel operations and precomputing gradient tables. Small scale simulations demonstrate reductions in average path length and boundary jitter compared to a pure edge tracking baseline, while maintaining environmental coverage measured through intersection merge analysis. Battery consumption and computational utilization confirm the feasibility of achieving 10, 15 m/s forward motion on standard micro platforms. This approach enables rapid deployment in the field, requiring robust sensing and minimal communications for emergency reconnaissance applications.
Authors: Peizheng Li, Ioannis Mavromatis, Ajith Sahadevan, Tim Farnham, Adnan Aijaz, Aftab Khan
Abstract: We present a large-scale, longitudinal visual dataset of urban streetlights captured by 22 fixed-angle cameras deployed across Bristol, U.K., from 2021 to 2025. The dataset contains over 526,000 images, collected hourly under diverse lighting, weather, and seasonal conditions. Each image is accompanied by rich metadata, including timestamps, GPS coordinates, and device identifiers. This unique real-world dataset enables detailed investigation of visual drift, anomaly detection, and MLOps strategies in smart city deployments. To promtoe seconardary analysis, we additionally provide a self-supervised framework based on convolutional variational autoencoders (CNN-VAEs). Models are trained separately for each camera node and for day/night image sets. We define two per-sample drift metrics: relative centroid drift, capturing latent space deviation from a baseline quarter, and relative reconstruction error, measuring normalized image-domain degradation. This dataset provides a realistic, fine-grained benchmark for evaluating long-term model stability, drift-aware learning, and deployment-ready vision systems. The images and structured metadata are publicly released in JPEG and CSV formats, supporting reproducibility and downstream applications such as streetlight monitoring, weather inference, and urban scene understanding. The dataset can be found at https://doi.org/10.5281/zenodo.17781192 and https://doi.org/10.5281/zenodo.17859120.
URLs: https://doi.org/10.5281/zenodo.17781192, https://doi.org/10.5281/zenodo.17859120.
Authors: Jeongjun Park, Sunwook Hwang, Hyeonho Noh, Jin Mo Yang, Hyun Jong Yang, Saewoong Bahk
Abstract: Distracted driving contributes to fatal crashes worldwide. To address this, researchers are using driver activity recognition (DAR) with impulse radio ultra-wideband (IR-UWB) radar, which offers advantages such as interference resistance, low power consumption, and privacy preservation. However, two challenges limit its adoption: the lack of large-scale real-world UWB datasets covering diverse distracted driving behaviors, and the difficulty of adapting fixed-input Vision Transformers (ViTs) to UWB radar data with non-standard dimensions. This work addresses both challenges. We present the ALERT dataset, which contains 10,220 radar samples of seven distracted driving activities collected in real driving conditions. We also propose the input-size-agnostic Vision Transformer (ISA-ViT), a framework designed for radar-based DAR. The proposed method resizes UWB data to meet ViT input requirements while preserving radar-specific information such as Doppler shifts and phase characteristics. By adjusting patch configurations and leveraging pre-trained positional embedding vectors (PEVs), ISA-ViT overcomes the limitations of naive resizing approaches. In addition, a domain fusion strategy combines range- and frequency-domain features to further improve classification performance. Comprehensive experiments demonstrate that ISA-ViT achieves a 22.68% accuracy improvement over an existing ViT-based approach for UWB-based DAR. By publicly releasing the ALERT dataset and detailing our input-size-agnostic strategy, this work facilitates the development of more robust and scalable distracted driving detection systems for real-world deployment.
Authors: Indranil Bhattacharjee, Vartika Narayani Srinet, Anirudha Bhattacharjee, Braj Bhushan, Bishakh Bhattacharya
Abstract: Understanding emotional responses in children with Autism Spectrum Disorder (ASD) during social interaction remains a critical challenge in both developmental psychology and human-robot interaction. This study presents a novel deep learning pipeline for emotion recognition in autistic children in response to a name-calling event by a humanoid robot (NAO), under controlled experimental settings. The dataset comprises of around 50,000 facial frames extracted from video recordings of 15 children with ASD. A hybrid model combining a fine-tuned ResNet-50-based Convolutional Neural Network (CNN) and a three-layer Graph Convolutional Network (GCN) trained on both visual and geometric features extracted from MediaPipe FaceMesh landmarks. Emotions were probabilistically labeled using a weighted ensemble of two models: DeepFace's and FER, each contributing to soft-label generation across seven emotion classes. Final classification leveraged a fused embedding optimized via Kullback-Leibler divergence. The proposed method demonstrates robust performance in modeling subtle affective responses and offers significant promise for affective profiling of ASD children in clinical and therapeutic human-robot interaction contexts, as the pipeline effectively captures micro emotional cues in neurodivergent children, addressing a major gap in autism-specific HRI research. This work represents the first such large-scale, real-world dataset and pipeline from India on autism-focused emotion analysis using social robotics, contributing an essential foundation for future personalized assistive technologies.
Authors: Zahra Dehghanian, Morteza Abolghasemi, Hamid Beigy, Hamid R. Rabiee
Abstract: Controllable video synthesis is a central challenge in computer vision, yet current models struggle with fine grained control beyond textual prompts, particularly for cinematic attributes like camera trajectory and genre. Existing datasets often suffer from severe data imbalance, noisy labels, or a significant simulation to real gap. To address this, we introduce CineLOG, a new dataset of 5,000 high quality, balanced, and uncut video clips. Each entry is annotated with a detailed scene description, explicit camera instructions based on a standard cinematic taxonomy, and genre label, ensuring balanced coverage across 17 diverse camera movements and 15 film genres. We also present our novel pipeline designed to create this dataset, which decouples the complex text to video (T2V) generation task into four easier stages with more mature technology. To enable coherent, multi shot sequences, we introduce a novel Trajectory Guided Transition Module that generates smooth spatio-temporal interpolation. Extensive human evaluations show that our pipeline significantly outperforms SOTA end to end T2V models in adhering to specific camera and screenplay instructions, while maintaining professional visual quality. All codes and data are available at https://cine-log.pages.dev.
Authors: Rheeya Uppaal, Phu Mon Htut, Min Bai, Nikolaos Pappas, Zheng Qi
Abstract: Reasoning-augmented vision language models (VLMs) generate explicit chains of thought that promise greater capability and transparency but also introduce new failure modes: models may reach correct answers via visually unfaithful intermediate steps, or reason faithfully yet fail on the final prediction. Standard evaluations that only measure final-answer accuracy cannot distinguish these behaviors. We introduce the visual faithfulness of reasoning chains as a distinct evaluation dimension, focusing on whether the perception steps of a reasoning chain are grounded in the image. We propose a training- and reference-free framework that decomposes chains into perception versus reasoning steps and uses off-the-shelf VLM judges for step-level faithfulness, additionally verifying this approach through a human meta-evaluation. Building on this metric, we present a lightweight self-reflection procedure that detects and locally regenerates unfaithful perception steps without any training. Across multiple reasoning-trained VLMs and perception-heavy benchmarks, our method reduces Unfaithful Perception Rate while preserving final-answer accuracy, improving the reliability of multimodal reasoning.
Authors: Zhi Chen, Jingcai Guo, Taotao Cai, Yuxiang Cai
Abstract: Recognizing unseen fine-grained categories demands a model that can distinguish subtle visual differences. This is typically achieved by transferring visual-attribute relationships from seen classes to unseen classes. The core challenge is attribute entanglement, where conventional models collapse distinct attributes like color, shape, and texture into a single visual embedding. This causes interference that masks these critical distinctions. The post-hoc solutions of previous work are insufficient, as they operate on representations that are already mixed. We propose a zero-shot learning framework that learns AttributeCentric Representations (ACR) to tackle this problem by imposing attribute disentanglement during representation learning. ACR is achieved with two mixture-of-experts components, including Mixture of Patch Experts (MoPE) and Mixture of Attribute Experts (MoAE). First, MoPE is inserted into the transformer using a dual-level routing mechanism to conditionally dispatch image patches to specialized experts. This ensures coherent attribute families are processed by dedicated experts. Finally, the MoAE head projects these expert-refined features into sparse, partaware attribute maps for robust zero-shot classification. On zero-shot learning benchmark datasets CUB, AwA2, and SUN, our ACR achieves consistent state-of-the-art results.
Authors: Minheng Ni, Zhengyuan Yang, Yaowen Zhang, Linjie Li, Chung-Ching Lin, Kevin Lin, Zhendong Wang, Xiaofei Wang, Shujie Liu, Lei Zhang, Wangmeng Zuo, Lijuan Wang
Abstract: We study professional image generation, where a model must synthesize information-dense, scientifically precise illustrations from technical descriptions rather than merely produce visually plausible pictures. To quantify the progress, we introduce ProImage-Bench, a rubric-based benchmark that targets biology schematics, engineering/patent drawings, and general scientific diagrams. For 654 figures collected from real textbooks and technical reports, we construct detailed image instructions and a hierarchy of rubrics that decompose correctness into 6,076 criteria and 44,131 binary checks. Rubrics are derived from surrounding text and reference figures using large multimodal models, and are evaluated by an automated LMM-based judge with a principled penalty scheme that aggregates sub-question outcomes into interpretable criterion scores. We benchmark several representative text-to-image models on ProImage-Bench and find that, despite strong open-domain performance, the best base model reaches only 0.791 rubric accuracy and 0.553 criterion score overall, revealing substantial gaps in fine-grained scientific fidelity. Finally, we show that the same rubrics provide actionable supervision: feeding failed checks back into an editing model for iterative refinement boosts a strong generator from 0.653 to 0.865 in rubric accuracy and from 0.388 to 0.697 in criterion score. ProImage-Bench thus offers both a rigorous diagnostic for professional image generation and a scalable signal for improving specification-faithful scientific illustrations.
Authors: Nathalie Alexander, Arnaud Gucciardi, Umberto Michelucci
Abstract: Accurate segmentation of infant brain MRI is essential for quantifying developmental changes in structure and complexity. However, ongoing myelination and reduced tissue contrast make automated segmentation particularly challenging. This study systematically compared segmentation accuracy and its impact on volumetric and fractal dimension (FD) estimates in infant brain MRI using the Baby Open Brains (BOB) dataset (71 scans, 1-9 months). Two methods, SynthSeg and SamSeg, were evaluated against expert annotations using Dice, Intersection over Union, 95th-percentile Hausdorff distance, and Normalised Mutual Information. SynthSeg outperformed SamSeg across all quality metrics (mean Dice > 0.8 for major regions) and provided volumetric estimates closely matching the manual reference (mean +4% [-28% - 71%]). SamSeg systematically overestimated ventricular and whole-brain volumes (mean +76% [-12% - 190%]). Segmentation accuracy improved with age, consistent with increasing tissue contrast during myelination. Fractal dimension a(FD) nalyses revealed significant regional differences between SynthSeg and expert segmentations, and Bland-Altman limits of agreement indicated that segmentation-related FD variability exceeded most group differences reported in developmental cohorts. Volume and FD deviations were positively correlated across structures, indicating that segmentation bias directly affects FD estimation. Overall, SynthSeg provided the most reliable volumetric and FD results for paediatric MRI, yet small morphological differences in volume and FD should be interpreted with caution due to segmentation-related uncertainty.
Authors: Tianyu Zhang, Dong Liu, Chang Wen Chen
Abstract: Ultra-low bitrate image compression (below 0.05 bits per pixel) is increasingly critical for bandwidth-constrained and computation-limited encoding scenarios such as edge devices. Existing frameworks typically rely on large pretrained encoders (e.g., VAEs or tokenizer-based models) and perform transform coding within their generative latent space. While these approaches achieve impressive perceptual fidelity, their reliance on heavy encoder networks makes them unsuitable for deployment on weak sender devices. In this work, we explore the feasibility of applying shallow encoders for ultra-low bitrate compression and propose a novel Asymmetric Extreme Image Compression (AEIC) framework that pursues simultaneously encoding simplicity and decoding quality. Specifically, AEIC employs moderate or even shallow encoder networks, while leveraging an one-step diffusion decoder to maintain high-fidelity and high-realism reconstructions under extreme bitrates. To further enhance the efficiency of shallow encoders, we design a dual-side feature distillation scheme that transfers knowledge from AEIC with moderate encoders to its shallow encoder variants. Experiments demonstrate that AEIC not only outperforms existing methods on rate-distortion-perception performance at ultra-low bitrates, but also delivers exceptional encoding efficiency for 35.8 FPS on 1080P input images, while maintaining competitive decoding speed compared to existing methods.
Authors: I Putu Andika Bagas Jiwanta, Ayu Purwarianti
Abstract: Detecting video moments and highlights from natural-language queries have been unified by transformer-based methods. Other works use generative Multimodal LLM (MLLM) to predict moments and/or highlights as text timestamps, utilizing its reasoning capability. While effective, text-based generation cannot provide direct gradients for frame-level predictions because the model only emits language tokens. Although recent Reinforcement Learning (RL) methods attempt to address the issue, we propose a novel approach by applying segmentation objectives directly on the LLM's output tokens. The LLM is fed with a fixed number of frames alongside a prompt that enforces it to output a sequence of continuous "0" and/or "1" characters, with one character per frame. The "0"/"1" characters benefit from the LLM's inherent language capability while also acting as background and foreground probabilities, respectively. Training employs segmentation losses on the probabilities alongside a normal causal LM loss. At inference, beam search generates sequence and logits, acting as moments and saliency scores, respectively. Despite sampling only 25 frames -- less than half of comparable methods -- our method achieved strong highlight detection (56.74 HIT@1) on QVHighlights. Additionally, our efficient method scores above the baseline (35.28 MAP) for moment retrieval. Empirically, segmentation losses provide a stable complementary learning signal even when the causal LM loss plateaus.
Authors: Yuqing Lei, Yingjun Du, Yawen Huang, Xiantong Zhen, Ling Shao
Abstract: Vision-language models (VLMs) such as CLIP exhibit strong zero-shot generalization but remain sensitive to domain shifts at test time. Test-time prompt tuning (TPT) mitigates this issue by adapting prompts with fixed augmentations, which may falter in more challenging settings. In this work, we propose Meta Test-Time Prompt Tuning (MetaTPT), a meta-learning framework that learns a self-supervised auxiliary task to guide test-time prompt tuning. The auxiliary task dynamically learns parameterized augmentations for each sample, enabling more expressive transformations that capture essential features in target domains. MetaTPT adopts a dual-loop optimization paradigm: an inner loop learns a self-supervised task that generates informative views, while the outer loop performs prompt tuning by enforcing consistency across these views. By coupling augmentation learning with prompt tuning, MetaTPT improves test-time adaptation under domain shifts. Extensive experiments demonstrate that MetaTPT achieves state-of-the-art performance on domain generalization and cross-dataset benchmarks.
Authors: Thibault Geoffroy, Myriam Maumy, Lionel Prevost
Abstract: As artificial intelligence (AI) systems become increasingly embedded in our daily life, the ability to recognize and adapt to human emotions is essential for effective human-computer interaction. Facial expression recognition (FER) provides a primary channel for inferring affective states, but the dynamic and culturally nuanced nature of emotions requires models that can learn continuously without forgetting prior knowledge. In this work, we propose a hybrid framework for FER in a continual learning setting that mitigates catastrophic forgetting. Our approach integrates two complementary modalities: deep convolutional features and facial Action Units (AUs) derived from the Facial Action Coding System (FACS). The combined representation is modelled through Bayesian Gaussian Mixture Models (BGMMs), which provide a lightweight, probabilistic solution that avoids retraining while offering strong discriminative power. Using the Compound Facial Expression of Emotion (CFEE) dataset, we show that our model can first learn basic expressions and then progressively recognize compound expressions. Experiments demonstrate improved accuracy, stronger knowledge retention, and reduced forgetting. This framework contributes to the development of emotionally intelligent AI systems with applications in education, healthcare, and adaptive user interfaces.
Authors: Jiahao Zhao
Abstract: Designing high-performance object detection architectures is a complex task, where traditional manual design is time-consuming and labor-intensive, and Neural Architecture Search (NAS) is computationally prohibitive. While recent approaches using Large Language Models (LLMs) show promise, they often function as iterative optimizers within a search loop, rather than generating architectures directly from a holistic understanding of the data. To address this gap, we propose Cognitive-YOLO, a novel framework for LLM-driven architecture synthesis that generates network configurations directly from the intrinsic characteristics of the dataset. Our method consists of three stages: first, an analysis module extracts key meta-features (e.g., object scale distribution and scene density) from the target dataset; second, the LLM reasons upon these features, augmented with state-of-the-art components retrieved via Retrieval-Augmented Generation (RAG), to synthesize the architecture into a structured Neural Architecture Description Language (NADL); finally, a compiler instantiates this description into a deployable model. Extensive experiments on five diverse object detection datasets demonstrate that our proposed Cognitive-YOLO consistently generates superior architectures, achieving highly competitive performance and demonstrating a superior performance-per-parameter trade-off compared to strong baseline models across multiple benchmarks. Crucially, our ablation studies prove that the LLM's data-driven reasoning is the primary driver of performance, demonstrating that a deep understanding of data "first principles" is more critical for achieving a superior architecture than simply retrieving SOTA components.
Authors: Ahmad Zafarani, Zahra Dehghanian, Mohammadreza Davoodi, Mohsen Shadroo, MohammadAmin Fazli, Hamid R. Rabiee
Abstract: The evaluation of drag based image editing models is unreliable due to a lack of standardized benchmarks and metrics. This ambiguity stems from inconsistent evaluation protocols and, critically, the absence of datasets containing ground truth target images, making objective comparisons between competing methods difficult. To address this, we introduce \textbf{RealDrag}, the first comprehensive benchmark for point based image editing that includes paired ground truth target images. Our dataset contains over 400 human annotated samples from diverse video sources, providing source/target images, handle/target points, editable region masks, and descriptive captions for both the image and the editing action. We also propose four novel, task specific metrics: Semantical Distance (SeD), Outer Mask Preserving Score (OMPS), Inner Patch Preserving Score (IPPS), and Directional Similarity (DiS). These metrics are designed to quantify pixel level matching fidelity, check preservation of non edited (out of mask) regions, and measure semantic alignment with the desired task. Using this benchmark, we conduct the first large scale systematic analysis of the field, evaluating 17 SOTA models. Our results reveal clear trade offs among current approaches and establish a robust, reproducible baseline to guide future research. Our dataset and evaluation toolkit will be made publicly available.
Authors: Hyunju Lee, Youngmin Oh, Jeimin Jeon, Donghyeon Baek, Bumsub Ham
Abstract: Transformer architecture search (TAS) aims to automatically discover efficient vision transformers (ViTs), reducing the need for manual design. Existing TAS methods typically train an over-parameterized network (i.e., a supernet) that encompasses all candidate architectures (i.e., subnets). However, all subnets share the same set of weights, which leads to interference that degrades the smaller subnets severely. We have found that well-trained small subnets can serve as a good foundation for training larger ones. Motivated by this, we propose a progressive training framework, dubbed GrowTAS, that begins with training small subnets and incorporate larger ones gradually. This enables reducing the interference and stabilizing a training process. We also introduce GrowTAS+ that fine-tunes a subset of weights only to further enhance the performance of large subnets. Extensive experiments on ImageNet and several transfer learning benchmarks, including CIFAR-10/100, Flowers, CARS, and INAT-19, demonstrate the effectiveness of our approach over current TAS methods
Authors: Huan Zheng, Yucheng Zhou, Tianyi Yan, Jiayi Su, Hongjun Chen, Dubing Chen, Wencheng Han, Runzhou Tao, Zhongying Qiu, Jianfei Yang, Jianbing Shen
Abstract: Current end-to-end autonomous driving systems operate at a level of intelligence akin to following simple steering commands. However, achieving genuinely intelligent autonomy requires a paradigm shift: moving from merely executing low-level instructions to understanding and fulfilling high-level, abstract human intentions. This leap from a command-follower to an intention-fulfiller, as illustrated in our conceptual framework, is hindered by a fundamental challenge: the absence of a standardized benchmark to measure and drive progress on this complex task. To address this critical gap, we introduce Intention-Drive, the first comprehensive benchmark designed to evaluate the ability to translate high-level human intent into safe and precise driving actions. Intention-Drive features two core contributions: (1) a new dataset of complex scenarios paired with corresponding natural language intentions, and (2) a novel evaluation protocol centered on the Intent Success Rate (ISR), which assesses the semantic fulfillment of the human's goal beyond simple geometric accuracy. Through an extensive evaluation of a spectrum of baseline models on Intention-Drive, we reveal a significant performance deficit, showing that the baseline model struggle to achieve the comprehensive scene and intention understanding required for this advanced task.
Authors: Yang Ou, Xiongwei Zhao, Xinye Yang, Yihan Wang, Yicheng Di, Rong Yuan, Xieyuanli Chen, Xu Zhu
Abstract: Unsupervised domain adaptation (UDA) enables semantic segmentation models to generalize from a labeled source domain to an unlabeled target domain. However, existing UDA methods still struggle to bridge the domain gap due to cross-domain contextual ambiguity, inconsistent feature representations, and class-wise pseudo-label noise. To address these challenges, we propose Omni-level Masking for Unsupervised Domain Adaptation (OMUDA), a unified framework that introduces hierarchical masking strategies across distinct representation levels. Specifically, OMUDA comprises: 1) a Context-Aware Masking (CAM) strategy that adaptively distinguishes foreground from background to balance global context and local details; 2) a Feature Distillation Masking (FDM) strategy that enhances robust and consistent feature learning through knowledge transfer from pre-trained models; and 3) a Class Decoupling Masking (CDM) strategy that mitigates the impact of noisy pseudo-labels by explicitly modeling class-wise uncertainty. This hierarchical masking paradigm effectively reduces the domain shift at the contextual, representational, and categorical levels, providing a unified solution beyond existing approaches. Extensive experiments on multiple challenging cross-domain semantic segmentation benchmarks validate the effectiveness of OMUDA. Notably, on the SYNTHIA->Cityscapes and GTA5->Cityscapes tasks, OMUDA can be seamlessly integrated into existing UDA methods and consistently achieving state-of-the-art results with an average improvement of 7%.
Authors: Benjamin Beilharz, Thomas S. A. Wallis
Abstract: While deep learning methods have achieved impressive success in many vision benchmarks, it remains difficult to understand and explain the representations and decisions of these models. Though vision models are typically trained on 2D inputs, they are often assumed to develop an implicit representation of the underlying 3D scene (for example, showing tolerance to partial occlusion, or the ability to reason about relative depth). Here, we introduce MRD (metamers rendered differentiably), an approach that uses physically based differentiable rendering to probe vision models' implicit understanding of generative 3D scene properties, by finding 3D scene parameters that are physically different but produce the same model activation (i.e. are model metamers). Unlike previous pixel-based methods for evaluating model representations, these reconstruction results are always grounded in physical scene descriptions. This means we can, for example, probe a model's sensitivity to object shape while holding material and lighting constant. As a proof-of-principle, we assess multiple models in their ability to recover scene parameters of geometry (shape) and bidirectional reflectance distribution function (material). The results show high similarity in model activation between target and optimized scenes, with varying visual results. Qualitatively, these reconstructions help investigate the physical scene attributes to which models are sensitive or invariant. MRD holds promise for advancing our understanding of both computer and human vision by enabling analysis of how physical scene parameters drive changes in model responses.
Authors: Shenghao Fu, Yukun Su, Fengyun Rao, Jing Lyu, Xiaohua Xie, Wei-Shi Zheng
Abstract: Open-vocabulary object detection aims to detect arbitrary classes via text prompts. Methods without cross-modal fusion layers (non-fusion) offer faster inference by treating recognition as a retrieval problem, \ie, matching regions to text queries in a shared embedding space. In this work, we fully explore this retrieval philosophy and demonstrate its unique advantages in efficiency and versatility through a model family named WeDetect: (1) State-of-the-art performance. WeDetect is a real-time detector with a dual-tower architecture. We show that, with well-curated data and full training, the non-fusion WeDetect surpasses other fusion models and establishes a strong open-vocabulary foundation. (2) Fast backtrack of historical data. WeDetect-Uni is a universal proposal generator based on WeDetect. We freeze the entire detector and only finetune an objectness prompt to retrieve generic object proposals across categories. Importantly, the proposal embeddings are class-specific and enable a new application, object retrieval, supporting retrieval objects in historical data. (3) Integration with LMMs for referring expression comprehension (REC). We further propose WeDetect-Ref, an LMM-based object classifier to handle complex referring expressions, which retrieves target objects from the proposal list extracted by WeDetect-Uni. It discards next-token prediction and classifies objects in a single forward pass. Together, the WeDetect family unifies detection, proposal generation, object retrieval, and REC under a coherent retrieval framework, achieving state-of-the-art performance across 15 benchmarks with high inference efficiency.
Authors: Maurya Goyal, Anuj Singh, Hadi Jamali-Rad
Abstract: Aligning diffusion model outputs with downstream objectives is essential for improving task-specific performance. Broadly, inference-time training-free approaches for aligning diffusion models can be categorized into two main strategies: sampling-based methods, which explore multiple candidate outputs and select those with higher reward signals, and gradient-guided methods, which use differentiable reward approximations to directly steer the generation process. In this work, we propose a universal algorithm, UniCoDe, which brings together the strengths of sampling and gradient-based guidance into a unified framework. UniCoDe integrates local gradient signals during sampling, thereby addressing the sampling inefficiency inherent in complex reward-based sampling approaches. By cohesively combining these two paradigms, UniCoDe enables more efficient sampling while offering better trade-offs between reward alignment and divergence from the diffusion unconditional prior. Empirical results demonstrate that UniCoDe remains competitive with state-of-the-art baselines across a range of tasks. The code is available at https://github.com/maurya-goyal10/UniCoDe
Authors: Zishen Song, Yongjian Zhu, Dong Wang, Hongzhan Liu, Lingyu Jiang, Yongxing Duan, Zehua Zhang, Sihan Li, Jiarui Li
Abstract: Timely and accurate detection of foliar diseases is vital for safeguarding crop growth and reducing yield losses. Yet, in real-field conditions, cluttered backgrounds, domain shifts, and limited lesion-level datasets hinder robust modeling. To address these challenges, we release Daylily-Leaf, a paired lesion-level dataset comprising 1,746 RGB images and 7,839 lesions captured under both ideal and in-field conditions, and propose TCLeaf-Net, a transformer-convolution hybrid detector optimized for real-field use. TCLeaf-Net is designed to tackle three major challenges. To mitigate interference from complex backgrounds, the transformer-convolution module (TCM) couples global context with locality-preserving convolution to suppress non-leaf regions. To reduce information loss during downsampling, the raw-scale feature recalling and sampling (RSFRS) block combines bilinear resampling and convolution to preserve fine spatial detail. To handle variations in lesion scale and feature shifts, the deformable alignment block with FPN (DFPN) employs offset-based alignment and multi-receptive-field perception to strengthen multi-scale fusion. Experimental results show that on the in-field split of the Daylily-Leaf dataset, TCLeaf-Net improves mAP@50 by 5.4 percentage points over the baseline model, reaching 78.2\%, while reducing computation by 7.5 GFLOPs and GPU memory usage by 8.7\%. Moreover, the model outperforms recent YOLO and RT-DETR series in both precision and recall, and demonstrates strong performance on the PlantDoc, Tomato-Leaf, and Rice-Leaf datasets, validating its robustness and generalizability to other plant disease detection scenarios.
Authors: Yufei Yin, Qianke Meng, Minghao Chen, Jiajun Ding, Zhenwei Shao, Zhou Yu
Abstract: Long-form video understanding remains challenging due to the extended temporal structure and dense multimodal cues. Despite recent progress, many existing approaches still rely on hand-crafted reasoning pipelines or employ token-consuming video preprocessing to guide MLLMs in autonomous reasoning. To overcome these limitations, we introduce VideoARM, an Agentic Reasoning-over-hierarchical-Memory paradigm for long-form video understanding. Instead of static, exhaustive preprocessing, VideoARM performs adaptive, on-the-fly agentic reasoning and memory construction. Specifically, VideoARM performs an adaptive and continuous loop of observing, thinking, acting, and memorizing, where a controller autonomously invokes tools to interpret the video in a coarse-to-fine manner, thereby substantially reducing token consumption. In parallel, a hierarchical multimodal memory continuously captures and updates multi-level clues throughout the operation of the agent, providing precise contextual information to support the controller in decision-making. Experiments on prevalent benchmarks demonstrate that VideoARM outperforms the state-of-the-art method, DVD, while significantly reducing token consumption for long-form videos.
Authors: Peixuan Zhang, Zijian Jia, Kaiqi Liu, Shuchen Weng, Si Li, Boxin Shi
Abstract: While recent advancements in generative models have achieved remarkable visual fidelity in video synthesis, creating coherent multi-shot narratives remains a significant challenge. To address this, keyframe-based approaches have emerged as a promising alternative to computationally intensive end-to-end methods, offering the advantages of fine-grained control and greater efficiency. However, these methods often fail to maintain cross-shot consistency and capture cinematic language. In this paper, we introduce STAGE, a SToryboard-Anchored GEneration workflow to reformulate the keyframe-based multi-shot video generation task. Instead of using sparse keyframes, we propose STEP2 to predict a structural storyboard composed of start-end frame pairs for each shot. We introduce the multi-shot memory pack to ensure long-range entity consistency, the dual-encoding strategy for intra-shot coherence, and the two-stage training scheme to learn cinematic inter-shot transition. We also contribute the large-scale ConStoryBoard dataset, including high-quality movie clips with fine-grained annotations for story progression, cinematic attributes, and human preferences. Extensive experiments demonstrate that STAGE achieves superior performance in structured narrative control and cross-shot coherence.
Authors: Hyunkoo Lee, Wooseok Jang, Jini Yang, Taehwan Kim, Sangoh Kim, Sangwon Jung, Seungryong Kim
Abstract: Video personalization aims to generate videos that faithfully reflect a user-provided subject while following a text prompt. However, existing approaches often rely on heavy video-based finetuning or large-scale video datasets, which impose substantial computational cost and are difficult to scale. Furthermore, they still struggle to maintain fine-grained appearance consistency across frames. To address these limitations, we introduce V-Warper, a training-free coarse-to-fine personalization framework for transformer-based video diffusion models. The framework enhances fine-grained identity fidelity without requiring any additional video training. (1) A lightweight coarse appearance adaptation stage leverages only a small set of reference images, which are already required for the task. This step encodes global subject identity through image-only LoRA and subject-embedding adaptation. (2) A inference-time fine appearance injection stage refines visual fidelity by computing semantic correspondences from RoPE-free mid-layer query--key features. These correspondences guide the warping of appearance-rich value representations into semantically aligned regions of the generation process, with masking ensuring spatial reliability. V-Warper significantly improves appearance fidelity while preserving prompt alignment and motion dynamics, and it achieves these gains efficiently without large-scale video finetuning.
Authors: Junqiao Fan, Yunjiao Zhou, Yizhuo Yang, Xinyuan Cui, Jiarui Zhang, Lihua Xie, Jianfei Yang, Chris Xiaoxuan Lu, Fangqiang Ding
Abstract: Human mesh reconstruction (HMR) provides direct insights into body-environment interaction, which enables various immersive applications. While existing large-scale HMR datasets rely heavily on line-of-sight RGB input, vision-based sensing is limited by occlusion, lighting variation, and privacy concerns. To overcome these limitations, recent efforts have explored radio-frequency (RF) mmWave radar for privacy-preserving indoor human sensing. However, current radar datasets are constrained by sparse skeleton labels, limited scale, and simple in-place actions. To advance the HMR research community, we introduce M4Human, the current largest-scale (661K-frame) ($9\times$ prior largest) multimodal benchmark, featuring high-resolution mmWave radar, RGB, and depth data. M4Human provides both raw radar tensors (RT) and processed radar point clouds (RPC) to enable research across different levels of RF signal granularity. M4Human includes high-quality motion capture (MoCap) annotations with 3D meshes and global trajectories, and spans 20 subjects and 50 diverse actions, including in-place, sit-in-place, and free-space sports or rehabilitation movements. We establish benchmarks on both RT and RPC modalities, as well as multimodal fusion with RGB-D modalities. Extensive results highlight the significance of M4Human for radar-based human modeling while revealing persistent challenges under fast, unconstrained motion. The dataset and code will be released after the paper publication.
Authors: Swayam Bhanded
Abstract: Recent advances have significantly improved the training efficiency of diffusion transformers. However, these techniques have largely been studied in isolation, leaving unexplored the potential synergies from combining multiple approaches. We present SR-DiT (Speedrun Diffusion Transformer), a framework that systematically integrates token routing, architectural improvements, and training modifications on top of representation alignment. Our approach achieves FID 3.49 and KDD 0.319 on ImageNet-256 using only a 140M parameter model at 400K iterations without classifier-free guidance - comparable to results from 685M parameter models trained significantly longer. To our knowledge, this is a state-of the-art result at this model size. Through extensive ablation studies, we identify which technique combinations are most effective and document both synergies and incompatibilities. We release our framework as a computationally accessible baseline for future research.
Authors: Haowen Wang, Xiaoping Yuan, Fugang Zhang, Rui Jian, Yuanwei Zhu, Xiuquan Qiao, Yakun Huang
Abstract: Generating articulated assets is crucial for robotics, digital twins, and embodied intelligence. Existing generative models often rely on single-view inputs representing closed states, resulting in ambiguous or unrealistic kinematic structures due to the entanglement between geometric shape and joint dynamics. To address these challenges, we introduce ArtGen, a conditional diffusion-based framework capable of generating articulated 3D objects with accurate geometry and coherent kinematics from single-view images or text descriptions at arbitrary part-level states. Specifically, ArtGen employs cross-state Monte Carlo sampling to explicitly enforce global kinematic consistency, reducing structural-motion entanglement. Additionally, we integrate a Chain-of-Thought reasoning module to infer robust structural priors, such as part semantics, joint types, and connectivity, guiding a sparse-expert Diffusion Transformer to specialize in diverse kinematic interactions. Furthermore, a compositional 3D-VAE latent prior enhanced with local-global attention effectively captures fine-grained geometry and global part-level relationships. Extensive experiments on the PartNet-Mobility benchmark demonstrate that ArtGen significantly outperforms state-of-the-art methods.
Authors: Khalfalla Awedat, Mohamed Abidalrekab, Mohammad El-Yabroudi
Abstract: Vertical beam dropout in spinning LiDAR sensors triggered by hardware aging, dust, snow, fog, or bright reflections removes entire vertical slices from the point cloud and severely degrades 3D perception in autonomous vehicles. This paper proposes a Graph Attention Network (GAT)-based framework that reconstructs these missing vertical channels using only the current LiDAR frame, with no camera images or temporal information required. Each LiDAR sweep is represented as an unstructured spatial graph: points are nodes and edges connect nearby points while preserving the original beam-index ordering. A multi-layer GAT learns adaptive attention weights over local geometric neighborhoods and directly regresses the missing elevation (z) values at dropout locations. Trained and evaluated on 1,065 raw KITTI sequences with simulated channel dropout, the method achieves an average height RMSE of 11.67 cm, with 87.98% of reconstructed points falling within a 10 cm error threshold. Inference takes 14.65 seconds per frame on a single GPU, and reconstruction quality remains stable for different neighborhood sizes k. These results show that a pure graph attention model operating solely on raw point-cloud geometry can effectively recover dropped vertical beams under realistic sensor degradation.
Authors: Tue-Thu Van-Dinh, Hoang-Duy Tran, Truong-Binh Duong, Mai-Hanh Pham, Binh-Nam Le-Nguyen, Quoc-Thai Nguyen
Abstract: Infographic Visual Question Answering (InfographicVQA) evaluates a model's ability to read and reason over data-rich, layout-heavy visuals that combine text, charts, icons, and design elements. Compared with scene-text or natural-image VQA, infographics require stronger integration of OCR, layout understanding, and numerical and semantic reasoning. We introduce ViInfographicVQA, the first benchmark for Vietnamese InfographicVQA, comprising over 6747 real-world infographics and 20409 human-verified question-answer pairs across economics, healthcare, education, and more. The benchmark includes two evaluation settings. The Single-image task follows the traditional setup in which each question is answered using a single infographic. The Multi-image task requires synthesizing evidence across multiple semantically related infographics and is, to our knowledge, the first Vietnamese evaluation of cross-image reasoning in VQA. We evaluate a range of recent vision-language models on this benchmark, revealing substantial performance disparities, with the most significant errors occurring on Multi-image questions that involve cross-image integration and non-span reasoning. ViInfographicVQA contributes benchmark results for Vietnamese InfographicVQA and sheds light on the limitations of current multimodal models in low-resource contexts, encouraging future exploration of layout-aware and cross-image reasoning methods.
Authors: Hangwei Zhang, Armando Teles Fortes, Tianyi Wei, Xingang Pan
Abstract: Bokeh and monocular depth estimation are tightly coupled through the same lens imaging geometry, yet current methods exploit this connection in incomplete ways. High-quality bokeh rendering pipelines typically depend on noisy depth maps, which amplify estimation errors into visible artifacts, while modern monocular metric depth models still struggle on weakly textured, distant and geometrically ambiguous regions where defocus cues are most informative. We introduce BokehDepth, a two-stage framework that decouples bokeh synthesis from depth prediction and treats defocus as an auxiliary supervision-free geometric cue. In Stage-1, a physically guided controllable bokeh generator, built on a powerful pretrained image editing backbone, produces depth-free bokeh stacks with calibrated bokeh strength from a single sharp input. In Stage-2, a lightweight defocus-aware aggregation module plugs into existing monocular depth encoders, fuses features along the defocus dimension, and exposes stable depth-sensitive variations while leaving downstream decoder unchanged. Across challenging benchmarks, BokehDepth improves visual fidelity over depth-map-based bokeh baselines and consistently boosts the metric accuracy and robustness of strong monocular depth foundation models.
Authors: Ke Zhang, Yiqun Mei, Jiacong Xu, Vishal M. Patel
Abstract: Producing long, coherent video sequences with stable 3D structure remains a major challenge, particularly in streaming scenarios. Motivated by this, we introduce Endless World, a real-time framework for infinite, 3D-consistent video generation.To support infinite video generation, we introduce a conditional autoregressive training strategy that aligns newly generated content with existing video frames. This design preserves long-range dependencies while remaining computationally efficient, enabling real-time inference on a single GPU without additional training overhead.Moreover, our Endless World integrates global 3D-aware attention to provide continuous geometric guidance across time. Our 3D injection mechanism enforces physical plausibility and geometric consistency throughout extended sequences, addressing key challenges in long-horizon and dynamic scene synthesis.Extensive experiments demonstrate that Endless World produces long, stable, and visually coherent videos, achieving competitive or superior performance to existing methods in both visual fidelity and spatial consistency. Our project has been available on https://bwgzk-keke.github.io/EndlessWorld/.
Authors: Jiachen Tao, Benjamin Planche, Van Nguyen Nguyen, Junyi Wu, Yuchun Liu, Haoxuan Wang, Zhongpai Gao, Gengyu Zhang, Meng Zheng, Feiran Wang, Anwesa Choudhuri, Zhenghao Zhao, Weitai Kang, Terrence Chen, Yan Yan, Ziyan Wu
Abstract: Accurately modeling light transport is essential for realistic image synthesis. Photon mapping provides physically grounded estimates of complex global illumination effects such as caustics and specular-diffuse interactions, yet its per-view radiance estimation remains computationally inefficient when rendering multiple views of the same scene. The inefficiency arises from independent photon tracing and stochastic kernel estimation at each viewpoint, leading to inevitable redundant computation. To accelerate multi-view rendering, we reformulate photon mapping as a continuous and reusable radiance function. Specifically, we introduce the Gaussian Photon Field (GPF), a learnable representation that encodes photon distributions as anisotropic 3D Gaussian primitives parameterized by position, rotation, scale, and spectrum. GPF is initialized from physically traced photons in the first SPPM iteration and optimized using multi-view supervision of final radiance, distilling photon-based light transport into a continuous field. Once trained, the field enables differentiable radiance evaluation along camera rays without repeated photon tracing or iterative refinement. Extensive experiments on scenes with complex light transport, such as caustics and specular-diffuse interactions, demonstrate that GPF attains photon-level accuracy while reducing computation by orders of magnitude, unifying the physical rigor of photon-based rendering with the efficiency of neural scene representations.
Authors: Hoang Anh Just, Yifei Fan, Handong Zhao, Jiuxiang Gu, Ruiyi Zhang, Simon Jenni, Kushal Kafle, Ruoxi Jia, Jing Shi
Abstract: Reinforcement learning from verifiable rewards (RLVR) has recently been extended from text-only LLMs to vision-language models (VLMs) to elicit long-chain multimodal reasoning. However, RLVR-trained VLMs still exhibit two persistent failure modes: inaccurate visual extraction (missing or hallucinating details) and logically inconsistent chains-of-thought, largely because verifiable signals supervise only the final answer. We propose PeRL-VL (Perception and Reasoning Learning for Vision-Language Models), a decoupled framework that separately improves visual perception and textual reasoning on top of RLVR. For perception, PeRL-VL introduces a VLM-based description reward that scores the model's self-generated image descriptions for faithfulness and sufficiency. For reasoning, PeRL-VL adds a text-only Reasoning SFT stage on logic-rich chain-of-thought data, enhancing coherence and logical consistency independently of vision. Across diverse multimodal benchmarks, PeRL-VL improves average Pass@1 accuracy from 63.3% (base Qwen2.5-VL-7B) to 68.8%, outperforming standard RLVR, text-only reasoning SFT, and naive multimodal distillation from GPT-4o.
Authors: Shengkai Xu, Hsiang Lun Kao, Tianxiang Xu, Honghui Zhang, Junqiao Wang, Runmeng Ding, Guanyu Liu, Tianyu Shi, Zhenyu Yu, Guofeng Pan, Ziqian Bi, Yuqi Ouyang
Abstract: Polyp detectors trained on clean datasets often underperform in real-world endoscopy, where illumination changes, motion blur, and occlusions degrade image quality. Existing approaches struggle with the domain gap between controlled laboratory conditions and clinical practice, where adverse imaging conditions are prevalent. In this work, we propose AdaptiveDetector, a novel two-stage detector-verifier framework comprising a YOLOv11 detector with a vision-language model (VLM) verifier. The detector adaptively adjusts per-frame confidence thresholds under VLM guidance, while the verifier is fine-tuned with Group Relative Policy Optimization (GRPO) using an asymmetric, cost-sensitive reward function specifically designed to discourage missed detections -- a critical clinical requirement. To enable realistic assessment under challenging conditions, we construct a comprehensive synthetic testbed by systematically degrading clean datasets with adverse conditions commonly encountered in clinical practice, providing a rigorous benchmark for zero-shot evaluation. Extensive zero-shot evaluation on synthetically degraded CVC-ClinicDB and Kvasir-SEG images demonstrates that our approach improves recall by 14 to 22 percentage points over YOLO alone, while precision remains within 0.7 points below to 1.7 points above the baseline. This combination of adaptive thresholding and cost-sensitive reinforcement learning achieves clinically aligned, open-world polyp detection with substantially fewer false negatives, thereby reducing the risk of missed precancerous polyps and improving patient outcomes.
Authors: Tasweer Ahmad, Arindam Sikdar, Sandip Pradhan, Ardhendu Behera
Abstract: Few-shot image classification remains difficult under limited supervision and visual domain shift. Recent cache-based adaptation approaches (e.g., Tip-Adapter) address this challenge to some extent by learning lightweight residual adapters over frozen features, yet they still inherit CLIP's tendency to encode global, general-purpose representations that are not optimally discriminative to adapt the generalist to the specialist's domain in low-data regimes. We address this limitation with a novel patch-driven relational refinement that learns cache adapter weights from intra-image patch dependencies rather than treating an image embedding as a monolithic vector. Specifically, we introduce a relational gated graph attention network that constructs a patch graph and performs edge-aware attention to emphasize informative inter-patch interactions, producing context-enriched patch embeddings. A learnable multi-aggregation pooling then composes these into compact, task-discriminative representations that better align cache keys with the target few-shot classes. Crucially, the proposed graph refinement is used only during training to distil relational structure into the cache, incurring no additional inference cost beyond standard cache lookup. Final predictions are obtained by a residual fusion of cache similarity scores with CLIP zero-shot logits. Extensive evaluations on 11 benchmarks show consistent gains over state-of-the-art CLIP adapter and cache-based baselines while preserving zero-shot efficiency. We further validate battlefield relevance by introducing an Injured vs. Uninjured Soldier dataset for casualty recognition. It is motivated by the operational need to support triage decisions within the "platinum minutes" and the broader "golden hour" window in time-critical UAV-driven search-and-rescue and combat casualty care.
Authors: Jinfan Zhou, Lixin Luo, Sungmin Eum, Heesung Kwon, Jeong Joon Park
Abstract: We explore spatiotemporal data augmentation using video foundation models to diversify both camera viewpoints and scene dynamics. Unlike existing approaches based on simple geometric transforms or appearance perturbations, our method leverages off-the-shelf video diffusion models to generate realistic 3D spatial and temporal variations from a given image dataset. Incorporating these synthesized video clips as supplemental training data yields consistent performance gains in low-data settings, such as UAV-captured imagery where annotations are scarce. Beyond empirical improvements, we provide practical guidelines for (i) choosing an appropriate spatiotemporal generative setup, (ii) transferring annotations to synthetic frames, and (iii) addressing disocclusion - regions newly revealed and unlabeled in generated views. Experiments on COCO subsets and UAV-captured datasets show that, when applied judiciously, spatiotemporal augmentation broadens the data distribution along axes underrepresented by traditional and prior generative methods, offering an effective lever for improving model performance in data-scarce regimes.
Authors: Qi Sun, Can Wang, Jiaxiang Shang, Wensen Feng, Jing Liao
Abstract: We present Animus3D, a text-driven 3D animation framework that generates motion field given a static 3D asset and text prompt. Previous methods mostly leverage the vanilla Score Distillation Sampling (SDS) objective to distill motion from pretrained text-to-video diffusion, leading to animations with minimal movement or noticeable jitter. To address this, our approach introduces a novel SDS alternative, Motion Score Distillation (MSD). Specifically, we introduce a LoRA-enhanced video diffusion model that defines a static source distribution rather than pure noise as in SDS, while another inversion-based noise estimation technique ensures appearance preservation when guiding motion. To further improve motion fidelity, we incorporate explicit temporal and spatial regularization terms that mitigate geometric distortions across time and space. Additionally, we propose a motion refinement module to upscale the temporal resolution and enhance fine-grained details, overcoming the fixed-resolution constraints of the underlying video model. Extensive experiments demonstrate that Animus3D successfully animates static 3D assets from diverse text prompts, generating significantly more substantial and detailed motion than state-of-the-art baselines while maintaining high visual integrity. Code will be released at https://qiisun.github.io/animus3d_page.
Authors: Huan Huang, Michele Esposito, Chen Zhao
Abstract: Accurate coronary artery segmentation from coronary computed tomography angiography is essential for quantitative coronary analysis and clinical decision support. Nevertheless, reliable segmentation remains challenging because of small vessel calibers, complex branching, blurred boundaries, and myocardial interference. We propose a coronary artery segmentation framework that integrates myocardial anatomical priors, structure aware feature encoding, and three dimensional wavelet inverse wavelet transformations. Myocardial priors and residual attention based feature enhancement are incorporated during encoding to strengthen coronary structure representation. Wavelet inverse wavelet based downsampling and upsampling enable joint spatial frequency modeling and preserve multi scale structural consistency, while a multi scale feature fusion module integrates semantic and geometric information in the decoding stage. The model is trained and evaluated on the public ImageCAS dataset using a 3D overlapping patch based strategy with a 7:1:2 split for training, validation, and testing. Experimental results demonstrate that the proposed method achieves a Dice coefficient of 0.8082, Sensitivity of 0.7946, Precision of 0.8471, and an HD95 of 9.77 mm, outperforming several mainstream segmentation models. Ablation studies further confirm the complementary contributions of individual components. The proposed method enables more stable and consistent coronary artery segmentation under complex geometric conditions, providing reliable segmentation results for subsequent coronary structure analysis tasks.
Authors: Shaif Chowdhury, Mushfika Rahman, Greg Hamerly
Abstract: We propose a supervised contrastive learning framework for video representation learning that leverages temporally global context. We introduce a video to image aggregation strategy that spatially arranges multiple frames from each video into a single input image. This design enables the use of pre trained convolutional neural network backbones such as ResNet50 and avoids the computational overhead of complex video transformer models. We then design a contrastive learning objective that directly compares pairwise projections generated by the model. Positive pairs are defined as projections from videos sharing the same label while all other projections are treated as negatives. Multiple natural views of the same video are created using different temporal frame samplings from the same underlying video. Rather than relying on data augmentation these frame level variations produce diverse positive samples with global context and reduce overfitting. Experiments on the Penn Action and HMDB51 datasets demonstrate that the proposed method outperforms existing approaches in classification accuracy while requiring fewer computational resources. The proposed Supervised Contrastive Frame Aggregation method learns effective video representations in both supervised and self supervised settings and supports video based tasks such as classification and captioning. The method achieves seventy six percent classification accuracy on Penn Action compared to forty three percent achieved by ViVIT and forty eight percent accuracy on HMDB51 compared to thirty seven percent achieved by ViVIT.
Authors: Xinqi Jin, Hanxun Yu, Bohan Yu, Kebin Liu, Jian Liu, Keda Tao, Yixuan Pei, Huan Wang, Fan Dang, Jiangchuan Liu, Weiqiang Wang
Abstract: Online video understanding is essential for applications like public surveillance and AI glasses. However, applying Multimodal Large Language Models (MLLMs) to this domain is challenging due to the large number of video frames, resulting in high GPU memory usage and computational latency. To address these challenges, we propose token pruning as a means to reduce context length while retaining critical information. Specifically, we introduce a novel redundancy metric, Maximum Similarity to Spatially Adjacent Video Tokens (MSSAVT), which accounts for both token similarity and spatial position. To mitigate the bidirectional dependency between pruning and redundancy, we further design a masked pruning strategy that ensures only mutually unadjacent tokens are pruned. We also integrate an existing temporal redundancy-based pruning method to eliminate temporal redundancy of the video modality. Experimental results on multiple online and offline video understanding benchmarks demonstrate that our method significantly improves the accuracy (i.e., by 4\% at most) while incurring a negligible pruning latency (i.e., less than 1ms). Our full implementation will be made publicly available.
Authors: Boyeong Im, Wooseok Lee, Yoojin Kwon, Hyung-Sin Kim
Abstract: To extend the application of vision-language models (VLMs) from web images to sensor-mediated physical environments, we propose Multi-View Physical-prompt for Test-Time Adaptation (MVP), a forward-only framework that moves test-time adaptation (TTA) from tokens to photons by treating the camera exposure triangle--ISO, shutter speed, and aperture--as physical prompts. At inference, MVP acquires a library of physical views per scene, selects the top-k sensor settings using a source-affinity score, evaluates each retained view under lightweight digital augmentations, filters the lowest-entropy subset of augmented views, and aggregates predictions with Zero-temperature softmax (i.e., hard voting). This selection-then-vote design is simple, calibration-friendly, and requires no gradients or model modifications. On ImageNet-ES and ImageNet-ES-Diverse, MVP consistently outperforms digital-only TTA on single Auto-Exposure captures, by up to 25.6 percentage points (pp), and delivers up to 3.4 pp additional gains over pipelines that combine conventional sensor control with TTA. MVP remains effective under reduced parameter candidate sets that lower capture latency, demonstrating practicality. These results support the main claim that, beyond post-capture prompting, measurement-time control--selecting and combining real physical views--substantially improves robustness for VLMs.
Authors: Lixin Chen, Chaomeng Chen, Jiale Zhou, Zhijian Wu, Xun Lin
Abstract: Despite the rapid progress of deep learning in video action recognition (VAR) in recent years, privacy leakage in videos remains a critical concern. Current state-of-the-art privacy-preserving methods often rely on anonymization. These methods suffer from (1) low concealment, where producing visually distorted videos that attract attackers' attention during transmission, and (2) spatiotemporal disruption, where degrading essential spatiotemporal features for accurate VAR. To address these issues, we propose StegaVAR, a novel framework that embeds action videos into ordinary cover videos and directly performs VAR in the steganographic domain for the first time. Throughout both data transmission and action analysis, the spatiotemporal information of hidden secret video remains complete, while the natural appearance of cover videos ensures the concealment of transmission. Considering the difficulty of steganographic domain analysis, we propose Secret Spatio-Temporal Promotion (STeP) and Cross-Band Difference Attention (CroDA) for analysis within the steganographic domain. STeP uses the secret video to guide spatiotemporal feature extraction in the steganographic domain during training. CroDA suppresses cover interference by capturing cross-band semantic differences. Experiments demonstrate that StegaVAR achieves superior VAR and privacy-preserving performance on widely used datasets. Moreover, our framework is effective for multiple steganographic models.
Authors: Indiwara Nanayakkara, Dehan Jayawickrama, Mervyn Parakrama B. Ekanayake
Abstract: Wire harness inspection process remains a labor-intensive process prone to errors in the modern Electronics Manufacturing Services (EMS) industry. This paper introduces a semiautomated machine vision system capable of verifying correct wire positioning, correctness of the connector polarity and correctness of color sequences for both linear and circular wire harness configurations. Five industrial standard CMOS cameras are integrated into a modularized mechanical framework in the physical structure of the solution and a HSV and RGB color domain value comparison based color sequence classifier is used in the operation. For each harness batch, a user can train the system using at least five reference samples; the trained file is stored and reused for similar harness types. The Solution is deployed at GPV Lanka Pvt. Ltd. (Fig. 2) and the system achieved 100% detection accuracy and reduced inspection time by 44% compared to manual methods. Additional features include user management, adjustable lighting, session data storage, and secure login. Results of this product usage in the real world situation demonstrate that this approach delivers reliable and efficient inspection capabilities.
Authors: Karthikeya KV
Abstract: This research introduces a transformative framework for integrating Vision-Enhanced Large Language Models (LLMs) with advanced transformer-based architectures to tackle challenges in high-resolution image synthesis and multimodal data interpretation. The proposed model incorporates a rectified flow mechanism that connects noise and data with linear paths, enabling efficient and high-quality generation. A bidirectional tokenization strategy is employed to seamlessly merge inputs from text, image, and video modalities, fostering a unified understanding across diverse data types. By embedding spatial-temporal features and leveraging a hybrid text-image sequence modeling approach, the framework achieves unparalleled fidelity in synthesized images and coherent multimodal representations. The architecture is optimized with a noise-aware learning algorithm, addressing discrepancies in noisy data distributions and improving generative performance under varying input conditions. Rigorous evaluations on benchmark datasets demonstrate a 25% increase in image resolution clarity and a 20% reduction in computational requirements compared to diffusion-based methods. Furthermore, the model exhibits robust scalability and adaptability, showcasing its potential in applications like autonomous systems, creative content generation, and advanced video analysis. This work underscores the role of vision-centric LLMs in redefining capabilities in computer vision and multimodal artificial intelligence.
Authors: Kei Yoshitake, Kento Hosono, Ken Kobayashi, Kazuhide Nakata
Abstract: In this paper, we propose a method for generating layouts for image-based advertisements by leveraging a Vision-Language Model (VLM). Conventional advertisement layout techniques have predominantly relied on saliency mapping to detect salient regions within a background image, but such approaches often fail to fully account for the image's detailed composition and semantic content. To overcome this limitation, our method harnesses a VLM to recognize the products and other elements depicted in the background and to inform the placement of text and logos. The proposed layout-generation pipeline consists of two steps. In the first step, the VLM analyzes the image to identify object types and their spatial relationships, then produces a text-based "placement plan" based on this analysis. In the second step, that plan is rendered into the final layout by generating HTML-format code. We validated the effectiveness of our approach through evaluation experiments, conducting both quantitative and qualitative comparisons against existing methods. The results demonstrate that by explicitly considering the background image's content, our method produces noticeably higher-quality advertisement layouts.
Authors: Cong Xie, Che Wang, Yan Zhang, Zheng Pan, Han Zou, Zhenpeng Zhan
Abstract: We study geometry-aware scene-consistent image generation: given a reference scene image and a text condition specifying an entity to be generated in the scene and its spatial relation to the scene, the goal is to synthesize an output image that preserves the same physical environment as the reference scene while correctly generating the entity according to the spatial relation described in the text. Existing methods struggle to balance scene preservation with prompt adherence: they either replicate the scene with high fidelity but poor responsiveness to the prompt, or prioritize prompt compliance at the expense of scene consistency. To resolve this trade-off, we introduce two key contributions: (i) a scene-consistent data construction pipeline that generates diverse, geometrically-grounded training pairs, and (ii) a novel geometry-guided attention loss that leverages cross-view cues to regularize the model's spatial reasoning. Experiments on our scene-consistent benchmark show that our approach achieves better scene alignment and text-image consistency than state-of-the-art baselines, according to both automatic metrics and human preference studies. Our method produces geometrically coherent images with diverse compositions that remain faithful to the textual instructions and the underlying scene structure.
Authors: Tingyan Wen, Haoyu Li, Yihuang Chen, Xing Zhou, Lifei Zhu, Xueqian Wang
Abstract: Diffusion models achieve remarkable generative quality, but computational overhead scales with step count, model depth, and sequence length. Feature caching is effective since adjacent timesteps yield highly similar features. However, an inherent trade-off remains: aggressive timestep reuse offers large speedups but can easily cross the critical line, hurting fidelity, while block- or token-level reuse is safer but yields limited computational savings. We present X-Slim (eXtreme-Slimming Caching), a training-free, cache-based accelerator that, to our knowledge, is the first unified framework to exploit cacheable redundancy across timesteps, structure (blocks), and space (tokens). Rather than simply mixing levels, X-Slim introduces a dual-threshold controller that turns caching into a push-then-polish process: it first pushes reuse at the timestep level up to an early-warning line, then switches to lightweight block- and token-level refresh to polish the remaining redundancy, and triggers full inference once the critical line is crossed to reset accumulated error. At each level, context-aware indicators decide when and where to cache. Across diverse tasks, X-Slim advances the speed-quality frontier. On FLUX.1-dev and HunyuanVideo, it reduces latency by up to 4.97x and 3.52x with minimal perceptual loss. On DiT-XL/2, it reaches 3.13x acceleration and improves FID by 2.42 over prior methods.
Authors: Wonseok Choi, Sohwi Lim, Nam Hyeon-Woo, Moon Ye-Bin, Dong-Ju Jeong, Jinyoung Hwang, Tae-Hyun Oh
Abstract: Instance-level image retrieval aims to find images containing the same object as a given query, despite variations in size, position, or appearance. To address this challenging task, we propose Patchify, a simple yet effective patch-wise retrieval framework that offers high performance, scalability, and interpretability without requiring fine-tuning. Patchify divides each database image into a small number of structured patches and performs retrieval by comparing these local features with a global query descriptor, enabling accurate and spatially grounded matching. To assess not just retrieval accuracy but also spatial correctness, we introduce LocScore, a localization-aware metric that quantifies whether the retrieved region aligns with the target object. This makes LocScore a valuable diagnostic tool for understanding and improving retrieval behavior. We conduct extensive experiments across multiple benchmarks, backbones, and region selection strategies, showing that Patchify outperforms global methods and complements state-of-the-art reranking pipelines. Furthermore, we apply Product Quantization for efficient large-scale retrieval and highlight the importance of using informative features during compression, which significantly boosts performance. Project website: https://wons20k.github.io/PatchwiseRetrieval/
Authors: Zihan Wang, Seungjun Lee, Guangzhao Dai, Gim Hee Lee
Abstract: Embodied agents face a critical dilemma that end-to-end models lack interpretability and explicit 3D reasoning, while modular systems ignore cross-component interdependencies and synergies. To bridge this gap, we propose the Dynamic 3D Vision-Language-Planning Model (D3D-VLP). Our model introduces two key innovations: 1) A Dynamic 3D Chain-of-Thought (3D CoT) that unifies planning, grounding, navigation, and question answering within a single 3D-VLM and CoT pipeline; 2) A Synergistic Learning from Fragmented Supervision (SLFS) strategy, which uses a masked autoregressive loss to learn from massive and partially-annotated hybrid data. This allows different CoT components to mutually reinforce and implicitly supervise each other. To this end, we construct a large-scale dataset with 10M hybrid samples from 5K real scans and 20K synthetic scenes that are compatible with online learning methods such as RL and DAgger. Our D3D-VLP achieves state-of-the-art results on multiple benchmarks, including Vision-and-Language Navigation (R2R-CE, REVERIE-CE, NavRAG-CE), Object-goal Navigation (HM3D-OVON), and Task-oriented Sequential Grounding and Navigation (SG3D). Real-world mobile manipulation experiments further validate the effectiveness.
Authors: Chengzhi Liu, Yuzhe Yang, Yue Fan, Qingyue Wei, Sheng Liu, Xin Eric Wang
Abstract: Recent advancements in Multimodal Large Language Models (MLLMs) have significantly enhanced cross-modal understanding and reasoning by incorporating Chain-of-Thought (CoT) reasoning in the semantic space. Building upon this, recent studies extend the CoT mechanism to the visual modality, enabling models to integrate visual information during reasoning through external tools or explicit image generation. However, these methods remain dependent on explicit step-by-step reasoning, unstable perception-reasoning interaction and notable computational overhead. Inspired by human cognition, we posit that thinking unfolds not linearly but through the dynamic interleaving of reasoning and perception within the mind. Motivated by this perspective, we propose DMLR, a test-time Dynamic Multimodal Latent Reasoning framework that employs confidence-guided latent policy gradient optimization to refine latent think tokens for in-depth reasoning. Furthermore, a Dynamic Visual Injection Strategy is introduced, which retrieves the most relevant visual features at each latent think token and updates the set of best visual patches. The updated patches are then injected into latent think token to achieve dynamic visual-textual interleaving. Experiments across seven multimodal reasoning benchmarks and various model architectures demonstrate that DMLR significantly improves reasoning and perception performance while maintaining high inference efficiency.
Authors: Zhou Tao, Shida Wang, Yongxiang Hua, Haoyu Cao, Linli Xu
Abstract: Multimodal Large Language Models have achieved impressive performance on a variety of vision-language tasks, yet their fine-grained visual perception and precise spatial reasoning remain limited. In this work, we introduce DiG (Differential Grounding), a novel proxy task framework where MLLMs learn fine-grained perception by identifying and localizing all differences between similar image pairs without prior knowledge of their number. To support scalable training, we develop an automated 3D rendering-based data generation pipeline that produces high-quality paired images with fully controllable discrepancies. To address the sparsity of difference signals, we further employ curriculum learning that progressively increases complexity from single to multiple differences, enabling stable optimization. Extensive experiments demonstrate that DiG significantly improves model performance across a variety of visual perception benchmarks and that the learned fine-grained perception skills transfer effectively to standard downstream tasks, including RefCOCO, RefCOCO+, RefCOCOg, and general multimodal perception benchmarks. Our results highlight differential grounding as a scalable and robust approach for advancing fine-grained visual reasoning in MLLMs.
Authors: Hongyang Li, Junyi Tao, Qijie Wei, Ningzhi Yang, Meng Wang, Weihong Yu, Xirong Li
Abstract: Previous work on cross-modal fundus image registration (CMFIR) assumes small cross-modal Field-of-View (FoV) disparity. By contrast, this paper is targeted at a more challenging scenario with large FoV disparity, to which directly applying current methods fails. We propose Crop and Alignment for cross-modal fundus image Registration(CARe), a very simple yet effective method. Specifically, given an OCTA with smaller FoV as a source image and a wide-field color fundus photograph (wfCFP) as a target image, our Crop operation exploits the physiological structure of the retina to crop from the target image a sub-image with its FoV roughly aligned with that of the source. This operation allows us to re-purpose the previous small-FoV-disparity oriented methods for subsequent image registration. Moreover, we improve spatial transformation by a double-fitting based Alignment module that utilizes the classical RANSAC algorithm and polynomial-based coordinate fitting in a sequential manner. Extensive experiments on a newly developed test set of 60 OCTA-wfCFP pairs verify the viability of CARe for CMFIR.
Authors: Qixin Xu, Haozhe Wang, Che Liu, Fangzhen Lin, Wenhu Chen
Abstract: Current document reasoning paradigms are constrained by a fundamental trade-off between scalability (processing long-context documents) and fidelity (capturing fine-grained, multimodal details). To bridge this gap, we propose CogDoc, a unified coarse-to-fine thinking framework that mimics human cognitive processes: a low-resolution "Fast Reading" phase for scalable information localization,followed by a high-resolution "Focused Thinking" phase for deep reasoning. We conduct a rigorous investigation into post-training strategies for the unified thinking framework, demonstrating that a Direct Reinforcement Learning (RL) approach outperforms RL with Supervised Fine-Tuning (SFT) initialization. Specifically, we find that direct RL avoids the "policy conflict" observed in SFT. Empirically, our 7B model achieves state-of-the-art performance within its parameter class, notably surpassing significantly larger proprietary models (e.g., GPT-4o) on challenging, visually rich document benchmarks.
Authors: Muhammad Umar Farooq, Abd Ur Rehman, Azka Rehman, Muhammad Usman, Dong-Kyu Chae, Junaid Qadir
Abstract: Accurate thyroid nodule segmentation in ultrasound images is critical for diagnosis and treatment planning. However, ambiguous boundaries between nodules and surrounding tissues, size variations, and the scarcity of annotated ultrasound data pose significant challenges for automated segmentation. Existing deep learning models struggle to incorporate contextual information from the thyroid gland and generalize effectively across diverse cases. To address these challenges, we propose SSMT-Net, a Semi-Supervised Multi-Task Transformer-based Network that leverages unlabeled data to enhance Transformer-centric encoder feature extraction capability in an initial unsupervised phase. In the supervised phase, the model jointly optimizes nodule segmentation, gland segmentation, and nodule size estimation, integrating both local and global contextual features. Extensive evaluations on the TN3K and DDTI datasets demonstrate that SSMT-Net outperforms state-of-the-art methods, with higher accuracy and robustness, indicating its potential for real-world clinical applications.
Authors: Sreehari Rajan, Kunal Bhosikar, Charu Sharma
Abstract: Generating realistic human motions that naturally respond to both spoken language and physical objects is crucial for interactive digital experiences. Current methods, however, address speech-driven gestures or object interactions independently, limiting real-world applicability due to a lack of integrated, comprehensive datasets. To overcome this, we introduce InteracTalker, a novel framework that seamlessly integrates prompt-based object-aware interactions with co-speech gesture generation. We achieve this by employing a multi-stage training process to learn a unified motion, speech, and prompt embedding space. To support this, we curate a rich human-object interaction dataset, formed by augmenting an existing text-to-motion dataset with detailed object interaction annotations. Our framework utilizes a Generalized Motion Adaptation Module that enables independent training, adapting to the corresponding motion condition, which is then dynamically combined during inference. To address the imbalance between heterogeneous conditioning signals, we propose an adaptive fusion strategy, which dynamically reweights the conditioning signals during diffusion sampling. InteracTalker successfully unifies these previously separate tasks, outperforming prior methods in both co-speech gesture generation and object-interaction synthesis, outperforming gesture-focused diffusion methods, yielding highly realistic, object-aware full-body motions with enhanced realism, flexibility, and control.
Authors: Haiyang Zheng, Nan Pu, Wenjing Li, Teng Long, Nicu Sebe, Zhun Zhong
Abstract: The proliferation of synthetic facial imagery has intensified the need for robust Open-World DeepFake Attribution (OW-DFA), which aims to attribute both known and unknown forgeries using labeled data for known types and unlabeled data containing a mixture of known and novel types. However, existing OW-DFA methods face two critical limitations: 1) A confidence skew that leads to unreliable pseudo-labels for novel forgeries, resulting in biased training. 2) An unrealistic assumption that the number of unknown forgery types is known *a priori*. To address these challenges, we propose a Confidence-Aware Asymmetric Learning (CAL) framework, which adaptively balances model confidence across known and novel forgery types. CAL mainly consists of two components: Confidence-Aware Consistency Regularization (CCR) and Asymmetric Confidence Reinforcement (ACR). CCR mitigates pseudo-label bias by dynamically scaling sample losses based on normalized confidence, gradually shifting the training focus from high- to low-confidence samples. ACR complements this by separately calibrating confidence for known and novel classes through selective learning on high-confidence samples, guided by their confidence gap. Together, CCR and ACR form a mutually reinforcing loop that significantly improves the model's OW-DFA performance. Moreover, we introduce a Dynamic Prototype Pruning (DPP) strategy that automatically estimates the number of novel forgery types in a coarse-to-fine manner, removing the need for unrealistic prior assumptions and enhancing the scalability of our methods to real-world OW-DFA scenarios. Extensive experiments on the standard OW-DFA benchmark and a newly extended benchmark incorporating advanced manipulations demonstrate that CAL consistently outperforms previous methods, achieving new state-of-the-art performance on both known and novel forgery attribution.
Authors: Yushun Tang, Ziqiong Liu, Jiyuan Jia, Yi Zhang, Zhihai He
Abstract: Online test-time adaptation aims to dynamically adjust a network model in real-time based on sequential input samples during the inference stage. In this work, we find that, when applying a transformer network model to a new target domain, the Query, Key, and Value features of its self-attention module often change significantly from those in the source domain, leading to substantial performance degradation of the transformer model. To address this important issue, we propose to develop a new approach to progressively recalibrate the self-attention at each layer using a local linear transform parameterized by conditioned scale and shift factors. We consider the online model adaptation from the source domain to the target domain as a progressive domain shift separation process. At each transformer network layer, we learn a Domain Separation Network to extract the domain shift feature, which is used to predict the scale and shift parameters for self-attention recalibration using a Factor Generator Network. These two lightweight networks are adapted online during inference. Experimental results on benchmark datasets demonstrate that the proposed progressive conditioned scale-shift recalibration (PCSR) method is able to significantly improve the online test-time domain adaptation performance by a large margin of up to 3.9\% in classification accuracy on the ImageNet-C dataset.
Authors: Yuran Wang, Bohan Zeng, Chengzhuo Tong, Wenxuan Liu, Yang Shi, Xiaochen Ma, Hao Liang, Yuanxing Zhang, Wentao Zhang
Abstract: Subject-driven image generation has advanced from single- to multi-subject composition, while neglecting distinction, the ability to identify and generate the correct subject when inputs contain multiple candidates. This limitation restricts effectiveness in complex, realistic visual settings. We propose Scone, a unified understanding-generation method that integrates composition and distinction. Scone enables the understanding expert to act as a semantic bridge, conveying semantic information and guiding the generation expert to preserve subject identity while minimizing interference. A two-stage training scheme first learns composition, then enhances distinction through semantic alignment and attention-based masking. We also introduce SconeEval, a benchmark for evaluating both composition and distinction across diverse scenarios. Experiments demonstrate that Scone outperforms existing open-source models in composition and distinction tasks on two benchmarks. Our model, benchmark, and training data are available at: https://github.com/Ryann-Ran/Scone.
Authors: Fatimah Zohra, Chen Zhao, Hani Itani, Bernard Ghanem
Abstract: CLIP achieves strong zero-shot image-text retrieval by aligning global vision and text representations, yet it falls behind on fine-grained tasks even when fine-tuned on long, detailed captions. In this work, we propose $\beta$-CLIP, a multi-granular text-conditioned contrastive learning framework designed to achieve hierarchical alignment between multiple textual granularities-from full captions to sentences and phrases-and their corresponding visual regions. For each level of granularity, $\beta$-CLIP utilizes cross-attention to dynamically pool image patches, producing contextualized visual embeddings. To address the semantic overlap inherent in this hierarchy, we introduce the $\beta$-Contextualized Contrastive Alignment Loss ($\beta$-CAL). This objective parameterizes the trade-off between strict query-specific matching and relaxed intra-image contextualization, supporting both soft Cross-Entropy and hard Binary Cross-Entropy formulations. Through extensive experiments, we demonstrate that $\beta$-CLIP significantly improves dense alignment: achieving 91.8% T2I 92.3% I2T at R@1 on Urban1K and 30.9% on FG-OVD (Hard), setting state-of-the-art among methods trained without hard negatives. $\beta$-CLIP establishes a robust, adaptive baseline for dense vision-language correspondence. The code and models are released at https://github.com/fzohra/B-CLIP.
Authors: Xue Li, Xiaonan Song, Henry Hu
Abstract: Real-world deployment of Vision-Language Models (VLMs) is hindered by high computational demands, as existing architectures inefficiently process all tokens uniformly. We introduce Adaptive Token Pruning (ATP), a dynamic inference mechanism that retains only the most informative tokens based on contextual relevance. ATP operates at the vision-language interface, assigning a hybrid importance score combining ViT CLS attention (intra-modal saliency) and CLIP text-image similarity (inter-modal relevance) to keep top-K tokens for the LLM. Unlike static compression, ATP adapts to each input without modifying the backbone. Proposed as a lightweight gating module, ATP is compatible with popular backbones like BLIP-2, LLaVA, and Flamingo. Preliminary evaluations across VQAv2, GQA, and COCO indicate that ATP reduces inference FLOPs by around 40% and achieves roughly 1.5x speedups in end-to-end latency with negligible accuracy loss (less than 1%). Qualitative analyses suggest ATP preserves visual grounding and enhances interpretability. Beyond efficiency, we investigate robustness under corruptions; observations suggest adaptive pruning suppresses spurious correlations, improving stability. These findings imply that resource-constrained inference and model reliability are not competing objectives. Finally, we discuss ATP's role in efficient multimodal edge computing pipelines.
Authors: Boyuan Li, Sipeng Zheng, Bin Cao, Ruihua Song, Zongqing Lu
Abstract: Extracting human motion from large-scale web videos offers a scalable solution to the data scarcity issue in character animation. However, some human parts in many video frames cannot be seen due to off-screen captures or occlusions. It brings a dilemma: discarding the data missing any part limits scale and diversity, while retaining it compromises data quality and model performance. To address this problem, we propose leveraging credible part-level data extracted from videos to enhance motion generation via a robust part-aware masked autoregression model. First, we decompose a human body into five parts and detect the parts clearly seen in a video frame as "credible". Second, the credible parts are encoded into latent tokens by our proposed part-aware variational autoencoder. Third, we propose a robust part-level masked generation model to predict masked credible parts, while ignoring those noisy parts. In addition, we contribute K700-M, a challenging new benchmark comprising approximately 200k real-world motion sequences, for evaluation. Experimental results indicate that our method successfully outperforms baselines on both clean and noisy datasets in terms of motion quality, semantic consistency and diversity. Project page: https://boyuaner.github.io/ropar-main/
Authors: Sehyun Kim, Hye Jun Lee, Jiwoo Lee, Changgyun Kim, Taemin Lee
Abstract: The spinal angle is an important indicator of body balance. It is important to restore the 3D shape of the human body and estimate the spine center line. Existing mul-ti-image-based body restoration methods require expensive equipment and complex pro-cedures, and single image-based body restoration methods have limitations in that it is difficult to accurately estimate the internal structure such as the spine center line due to occlusion and viewpoint limitation. This study proposes a method to compensate for the shortcomings of the multi-image-based method and to solve the limitations of the sin-gle-image method. We propose a 3D body posture analysis system that integrates depth images from four directions to restore a 3D human model and automatically estimate the spine center line. Through hierarchical matching of global and fine registration, restora-tion to noise and occlusion is performed. Also, the Adaptive Vertex Reduction is applied to maintain the resolution and shape reliability of the mesh, and the accuracy and stabil-ity of spinal angle estimation are simultaneously secured by using the Level of Detail en-semble. The proposed method achieves high-precision 3D spine registration estimation without relying on training data or complex neural network models, and the verification confirms the improvement of matching quality.
Authors: Zhenya Yang, Zhe Liu, Yuxiang Lu, Liping Hou, Chenxuan Miao, Siyi Peng, Bailan Feng, Xiang Bai, Hengshuang Zhao
Abstract: Physics-aware driving world model is essential for drive planning, out-of-distribution data synthesis, and closed-loop evaluation. However, existing methods often rely on a single diffusion model to directly map driving actions to videos, which makes learning difficult and leads to physically inconsistent outputs. To overcome these challenges, we propose GenieDrive, a novel framework designed for physics-aware driving video generation. Our approach starts by generating 4D occupancy, which serves as a physics-informed foundation for subsequent video generation. 4D occupancy contains rich physical information, including high-resolution 3D structures and dynamics. To facilitate effective compression of such high-resolution occupancy, we propose a VAE that encodes occupancy into a latent tri-plane representation, reducing the latent size to only 58% of that used in previous methods. We further introduce Mutual Control Attention (MCA) to accurately model the influence of control on occupancy evolution, and we jointly train the VAE and the subsequent prediction module in an end-to-end manner to maximize forecasting accuracy. Together, these designs yield a 7.2% improvement in forecasting mIoU at an inference speed of 41 FPS, while using only 3.47 M parameters. Additionally, a Normalized Multi-View Attention is introduced in the video generation model to generate multi-view driving videos with guidance from our 4D occupancy, significantly improving video quality with a 20.7% reduction in FVD. Experiments demonstrate that GenieDrive enables highly controllable, multi-view consistent, and physics-aware driving video generation.
Authors: Yue Jiang, Dingkang Yang, Minghao Han, Jinghang Han, Zizhi Chen, Yizhou Liu, Mingcheng Li, Peng Zhai, Lihua Zhang
Abstract: Despite rapid progress in multimodal large language models (MLLMs) and emerging omni-modal architectures, current benchmarks remain limited in scope and integration, suffering from incomplete modality coverage, restricted interaction to text-centric outputs, and weak interdependence and complementarity among modalities. To bridge these gaps, we introduce FysicsWorld, the first unified full-modality benchmark that supports bidirectional input-output across image, video, audio, and text, enabling comprehensive any-to-any evaluation across understanding, generation, and reasoning. FysicsWorld encompasses 16 primary tasks and 3,268 curated samples, aggregated from over 40 high-quality sources and covering a rich set of open-domain categories with diverse question types. We also propose the Cross-Modal Complementarity Screening (CMCS) strategy integrated in a systematic data construction framework that produces omni-modal data for spoken interaction and fusion-dependent cross-modal reasoning. Through a comprehensive evaluation of over 30 state-of-the-art baselines, spanning MLLMs, modality-specific models, unified understanding-generation models, and omni-modal language models, FysicsWorld exposes the performance disparities and limitations across models in understanding, generation, and reasoning. Our benchmark establishes a unified foundation and strong baselines for evaluating and advancing next-generation full-modality architectures.
Authors: Tianjiao Yu, Xinzhuo Li, Yifan Shen, Yuanzhe Liu, Ismini Lourentzou
Abstract: Recent advances in large multimodal models suggest that explicit reasoning mechanisms play a critical role in improving model reliability, interpretability, and cross-modal alignment. While such reasoning-centric approaches have been proven effective in language and vision tasks, their extension to 3D remains underdeveloped. CoRe3D introduces a unified 3D understanding and generation reasoning framework that jointly operates over semantic and spatial abstractions, enabling high-level intent inferred from language to directly guide low-level 3D content formation. Central to this design is a spatially grounded reasoning representation that decomposes 3D latent space into localized regions, allowing the model to reason over geometry in a compositional and procedural manner. By tightly coupling semantic chain-of-thought inference with structured spatial reasoning, CoRe3D produces 3D outputs that exhibit strong local consistency and faithful alignment with linguistic descriptions.
Authors: Hao Wang, Ashish Bastola, Chaoyi Zhou, Wenhui Zhu, Xiwen Chen, Xuanzhao Dong, Siyu Huang, Abolfazl Razi
Abstract: As generative models become increasingly capable of producing high-fidelity visual content, the demand for efficient, interpretable, and editable image representations has grown substantially. Recent advances in 2D Gaussian Splatting (2DGS) have emerged as a promising solution, offering explicit control, high interpretability, and real-time rendering capabilities (>1000 FPS). However, high-quality 2DGS typically requires post-optimization. Existing methods adopt random or heuristics (e.g., gradient maps), which are often insensitive to image complexity and lead to slow convergence (>10s). More recent approaches introduce learnable networks to predict initial Gaussian configurations, but at the cost of increased computational and architectural complexity. To bridge this gap, we present Fast-2DGS, a lightweight framework for efficient Gaussian image representation. Specifically, we introduce Deep Gaussian Prior, implemented as a conditional network to capture the spatial distribution of Gaussian primitives under different complexities. In addition, we propose an attribute regression network to predict dense Gaussian properties. Experiments demonstrate that this disentangled architecture achieves high-quality reconstruction in a single forward pass, followed by minimal fine-tuning. More importantly, our approach significantly reduces computational cost without compromising visual quality, bringing 2DGS closer to industry-ready deployment.
Authors: Tiange Zhang, Zhimeng Huang, Xiandong Meng, Kai Zhang, Zhipin Deng, Siwei Ma
Abstract: Neural Video Compression has emerged in recent years, with condition-based frameworks outperforming traditional codecs. However, most existing methods rely solely on the previous frame's features to predict temporal context, leading to two critical issues. First, the short reference window misses long-term dependencies and fine texture details. Second, propagating only feature-level information accumulates errors over frames, causing prediction inaccuracies and loss of subtle textures. To address these, we propose the Long-term Spatio-Temporal Enhanced Context (L-STEC) method. We first extend the reference chain with LSTM to capture long-term dependencies. We then incorporate warped spatial context from the pixel domain, fusing spatio-temporal information through a multi-receptive field network to better preserve reference details. Experimental results show that L-STEC significantly improves compression by enriching contextual information, achieving 37.01% bitrate savings in PSNR and 31.65% in MS-SSIM compared to DCVC-TCM, outperforming both VTM-17.0 and DCVC-FM and establishing new state-of-the-art performance.
Authors: Zhe Liu, Runhui Huang, Rui Yang, Siming Yan, Zining Wang, Lu Hou, Di Lin, Xiang Bai, Hengshuang Zhao
Abstract: Although multi-modal large language models (MLLMs) have shown strong capabilities across diverse domains, their application in generating fine-grained 3D perception and prediction outputs in autonomous driving remains underexplored. In this paper, we propose DrivePI, a novel spatial-aware 4D MLLM that serves as a unified Vision-Language-Action (VLA) framework that is also compatible with vision-action (VA) models. Our method jointly performs spatial understanding, 3D perception (i.e., 3D occupancy), prediction (i.e., occupancy flow), and planning (i.e., action outputs) in parallel through end-to-end optimization. To obtain both precise geometric information and rich visual appearance, our approach integrates point clouds, multi-view images, and language instructions within a unified MLLM architecture. We further develop a data engine to generate text-occupancy and text-flow QA pairs for 4D spatial understanding. Remarkably, with only a 0.5B Qwen2.5 model as MLLM backbone, DrivePI as a single unified model matches or exceeds both existing VLA models and specialized VA models. Specifically, compared to VLA models, DrivePI outperforms OpenDriveVLA-7B by 2.5% mean accuracy on nuScenes-QA and reduces collision rate by 70% over ORION (from 0.37% to 0.11%) on nuScenes. Against specialized VA models, DrivePI surpasses FB-OCC by 10.3 RayIoU for 3D occupancy on OpenOcc, reduces the mAVE from 0.591 to 0.509 for occupancy flow on OpenOcc, and achieves 32% lower L2 error than VAD (from 0.72m to 0.49m) for planning on nuScenes. Code will be available at https://github.com/happinesslz/DrivePI
Authors: Yunlong He, Gwilherm Lesn\'e, Ziqian Liu, Micha\"el Soumm, Pietro Gori
Abstract: Recent advancements in image synthesis have enabled high-quality image generation and manipulation. Most works focus on: 1) conditional manipulation, where an image is modified conditioned on a given attribute, or 2) disentangled representation learning, where each latent direction should represent a distinct semantic attribute. In this paper, we focus on a different and less studied research problem, called Contrastive Analysis (CA). Given two image datasets, we want to separate the common generative factors, shared across the two datasets, from the salient ones, specific to only one dataset. Compared to existing methods, which use attributes as supervised signals for editing (e.g., glasses, gender), the proposed method is weaker, since it only uses the dataset signal. We propose a novel framework for CA, that can be adapted to both GAN and Diffusion models, to learn both common and salient factors. By defining new and well-adapted learning strategies and losses, we ensure a relevant separation between common and salient factors, preserving a high-quality generation. We evaluate our approach on diverse datasets, covering human faces, animal images and medical scans. Our framework demonstrates superior separation ability and image quality synthesis compared to prior methods.
Authors: Yongyuan Liang, Xiyao Wang, Yuanchen Ju, Jianwei Yang, Furong Huang
Abstract: Scaling large multimodal models (LMMs) to 3D understanding poses unique challenges: point cloud data is sparse and irregular, existing models rely on fragmented architectures with modality-specific encoders, and training pipelines often suffer from instability and poor scalability. We introduce Lemon, a unified transformer architecture that addresses these challenges by jointly processing 3D point cloud patches and language tokens as a single sequence. Unlike prior work that relies on modality-specific encoders and cross-modal alignment modules, this design enables early spatial-linguistic fusion, eliminates redundant encoders, improves parameter efficiency, and supports more effective model scaling. To handle the complexity of 3D data, we develop a structured patchification and tokenization scheme that preserves spatial context, and a three-stage training curriculum that progressively builds capabilities from object-level recognition to scene-level spatial reasoning. Lemon establishes new state-of-the-art performance across comprehensive 3D understanding and reasoning tasks, from object recognition and captioning to spatial reasoning in 3D scenes, while demonstrating robust scaling properties as model size and training data increase. Our work provides a unified foundation for advancing 3D spatial intelligence in real-world applications.
Authors: N. K. B. M. P. K. B. Narasinghe, Uthayasanker Thayasivam
Abstract: Large-scale multimodal foundation models, particularly Contrastive Captioners (CoCa), have achieved state-of-the-art results by unifying contrastive alignment with generative captioning. While zero-shot transfer capabilities are well-documented, the adaptation of these generative-contrastive hybrids to downstream tasks with extreme data scarcity (few-shot learning) remains under-explored. Existing literature predominantly focuses on dual-encoder architectures like CLIP, leaving a gap in understanding how CoCa's distinct latent space responds to parameter-efficient fine-tuning (PEFT). This paper presents a comprehensive empirical study on adapting the CoCa visual backbone for few-shot image classification. We systematically evaluate a hierarchy of strategies, ranging from training-free hybrid prototyping to deep parameter adaptation via Low-Rank Adaptation (LoRA). First, we identify an "augmentation divergence": while strong data augmentation degrades the performance of linear probing in low-shot settings, it is essential for stabilizing LoRA fine-tuning. We also demonstrate that hybrid objectives incorporating Supervised Contrastive (SupCon) loss yield consistent performance improvements over standard Cross-Entropy across varying shot counts. Crucially, we characterize the sensitivity of training configurations to data scarcity, providing empirical reference settings for scaling regularization, rank, and sampling strategies to facilitate the efficient adaptation of generative-contrastive foundation models.
Authors: Weihan Xu, Kan Jen Cheng, Koichi Saito, Muhammad Jehanzeb Mirza, Tingle Li, Yisi Liu, Alexander H. Liu, Liming Wang, Masato Ishii, Takashi Shibuya, Yuki Mitsufuji, Gopala Anumanchipalli, Paul Pu Liang
Abstract: Joint editing of audio and visual content is crucial for precise and controllable content creation. This new task poses challenges due to the limitations of paired audio-visual data before and after targeted edits, and the heterogeneity across modalities. To address the data and modeling challenges in joint audio-visual editing, we introduce SAVEBench, a paired audiovisual dataset with text and mask conditions to enable object-grounded source-to-target learning. With SAVEBench, we train the Schrodinger Audio-Visual Editor (SAVE), an end-to-end flow-matching model that edits audio and video in parallel while keeping them aligned throughout processing. SAVE incorporates a Schrodinger Bridge that learns a direct transport from source to target audiovisual mixtures. Our evaluation demonstrates that the proposed SAVE model is able to remove the target objects in audio and visual content while preserving the remaining content, with stronger temporal synchronization and audiovisual semantic correspondence compared with pairwise combinations of an audio editor and a video editor.
Authors: Xiangzhong Liu, Jiajie Zhang, Hao Shen
Abstract: In automotive sensor fusion systems, smart sensors and Vehicle-to-Everything (V2X) modules are commonly utilized. Sensor data from these systems are typically available only as processed object lists rather than raw sensor data from traditional sensors. Instead of processing other raw data separately and then fusing them at the object level, we propose an end-to-end cross-level fusion concept with Transformer, which integrates highly abstract object list information with raw camera images for 3D object detection. Object lists are fed into a Transformer as denoising queries and propagated together with learnable queries through the latter feature aggregation process. Additionally, a deformable Gaussian mask, derived from the positional and size dimensional priors from the object lists, is explicitly integrated into the Transformer decoder. This directs attention toward the target area of interest and accelerates model training convergence. Furthermore, as there is no public dataset containing object lists as a standalone modality, we propose an approach to generate pseudo object lists from ground-truth bounding boxes by simulating state noise and false positives and negatives. As the first work to conduct cross-level fusion, our approach shows substantial performance improvements over the vision-based baseline on the nuScenes dataset. It demonstrates its generalization capability over diverse noise levels of simulated object lists and real detectors.
Authors: Minghao Zhu, Zhihao Zhang, Anmol Sidhu, Keith Redmill
Abstract: Automated road sign recognition is a critical task for intelligent transportation systems, but traditional deep learning methods struggle with the sheer number of sign classes and the impracticality of creating exhaustive labeled datasets. This paper introduces a novel zero-shot recognition framework that adapts the Retrieval-Augmented Generation (RAG) paradigm to address this challenge. Our method first uses a Vision Language Model (VLM) to generate a textual description of a sign from an input image. This description is used to retrieve a small set of the most relevant sign candidates from a vector database of reference designs. Subsequently, a Large Language Model (LLM) reasons over the retrieved candidates to make a final, fine-grained recognition. We validate this approach on a comprehensive set of 303 regulatory signs from the Ohio MUTCD. Experimental results demonstrate the framework's effectiveness, achieving 95.58% accuracy on ideal reference images and 82.45% on challenging real-world road data. This work demonstrates the viability of RAG-based architectures for creating scalable and accurate systems for road sign recognition without task-specific training.
Authors: Han Liu, Bogdan Georgescu, Yanbo Zhang, Youngjin Yoo, Michael Baumgartner, Riqiang Gao, Jianing Wang, Gengyan Zhao, Eli Gibson, Dorin Comaniciu, Sasa Grbic
Abstract: 3D medical image classification is essential for modern clinical workflows. Medical foundation models (FMs) have emerged as a promising approach for scaling to new tasks, yet current research suffers from three critical pitfalls: data-regime bias, suboptimal adaptation, and insufficient task coverage. In this paper, we address these pitfalls and introduce AnyMC3D, a scalable 3D classifier adapted from 2D FMs. Our method scales efficiently to new tasks by adding only lightweight plugins (about 1M parameters per task) on top of a single frozen backbone. This versatile framework also supports multi-view inputs, auxiliary pixel-level supervision, and interpretable heatmap generation. We establish a comprehensive benchmark of 12 tasks covering diverse pathologies, anatomies, and modalities, and systematically analyze state-of-the-art 3D classification techniques. Our analysis reveals key insights: (1) effective adaptation is essential to unlock FM potential, (2) general-purpose FMs can match medical-specific FMs if properly adapted, and (3) 2D-based methods surpass 3D architectures for 3D classification. For the first time, we demonstrate the feasibility of achieving state-of-the-art performance across diverse applications using a single scalable framework (including 1st place in the VLM3D challenge), eliminating the need for separate task-specific models.
Authors: Abhinav Kumar, Tristan Aumentado-Armstrong, Lazar Valkov, Gopal Sharma, Alex Levinshtein, Radek Grzeszczuk, Suren Kumar
Abstract: Accurately learning high-frequency signals is a challenge in computer vision and graphics, as neural networks often struggle with these signals due to spectral bias or optimization difficulties. While current techniques like Fourier encodings have made great strides in improving performance, there remains scope for improvement when presented with high-frequency information. This paper introduces Queried-Convolutions (Qonvolutions), a simple yet powerful modification using the neighborhood properties of convolution. Qonvolution convolves a low-frequency signal with queries (such as coordinates) to enhance the learning of intricate high-frequency signals. We empirically demonstrate that Qonvolutions enhance performance across a variety of high-frequency learning tasks crucial to both the computer vision and graphics communities, including 1D regression, 2D super-resolution, 2D image regression, and novel view synthesis (NVS). In particular, by combining Gaussian splatting with Qonvolutions for NVS, we showcase state-of-the-art performance on real-world complex scenes, even outperforming powerful radiance field models on image quality.
Authors: Zhimao Peng, Enguang Wang, Xialei Liu, Ming-Ming Cheng
Abstract: Semantically coherent out-of-distribution detection (SCOOD) is a recently proposed realistic OOD detection setting: given labeled in-distribution (ID) data and mixed in-distribution and out-of-distribution unlabeled data as the training data, SCOOD aims to enable the trained model to accurately identify OOD samples in the testing data. Current SCOOD methods mainly adopt various clustering-based in-distribution sample filtering (IDF) strategies to select clean ID samples from unlabeled data, and take the remaining samples as auxiliary OOD data, which inevitably introduces a large number of noisy samples in training. To address the above issue, we propose a concise SCOOD framework based on predictive sample assignment (PSA). PSA includes a dual-threshold ternary sample assignment strategy based on the predictive energy score that can significantly improve the purity of the selected ID and OOD sample sets by assigning unconfident unlabeled data to an additional discard sample set, and a concept contrastive representation learning loss to further expand the distance between ID and OOD samples in the representation space to assist ID/OOD discrimination. In addition, we also introduce a retraining strategy to help the model fully fit the selected auxiliary ID/OOD samples. Experiments on two standard SCOOD benchmarks demonstrate that our approach outperforms the state-of-the-art methods by a significant margin.
Authors: Zhimao Peng, Enguang Wang, Fei Yang, Xialei Liu, Ming-Ming Cheng
Abstract: Generalized category discovery (GCD) is an important and challenging task in open-world learning. Specifically, given some labeled data of known classes, GCD aims to cluster unlabeled data that contain both known and unknown classes. Current GCD methods based on parametric classification adopt the DINO-like pseudo-labeling strategy, where the sharpened probability output of one view is used as supervision information for the other view. However, large pre-trained models have a preference for some specific visual patterns, resulting in encoding spurious correlation for unlabeled data and generating noisy pseudo-labels. To address this issue, we propose a novel method, which contains two modules: Loss Sharpness Penalty (LSP) and Dynamic Anchor Selection (DAS). LSP enhances the robustness of model parameters to small perturbations by minimizing the worst-case loss sharpness of the model, which suppressing the encoding of trivial features, thereby reducing overfitting of noise samples and improving the quality of pseudo-labels. Meanwhile, DAS selects representative samples for the unknown classes based on KNN density and class probability during the model training and assigns hard pseudo-labels to them, which not only alleviates the confidence difference between known and unknown classes but also enables the model to quickly learn more accurate feature distribution for the unknown classes, thus further improving the clustering accuracy. Extensive experiments demonstrate that the proposed method can effectively mitigate the noise of pseudo-labels, and achieve state-of-the-art results on multiple GCD benchmarks.
Authors: Huu-An Vu, Van-Khanh Mai, Trong-Tam Nguyen, Quang-Duc Dam, Tien-Huy Nguyen, Thanh-Huong Le
Abstract: The rapid expansion of video content across online platforms has accelerated the need for retrieval systems capable of understanding not only isolated visual moments but also the temporal structure of complex events. Existing approaches often fall short in modeling temporal dependencies across multiple events and in handling queries that reference unseen or rare visual concepts. To address these challenges, we introduce MADTempo, a video retrieval framework developed by our team, AIO_Trinh, that unifies temporal search with web-scale visual grounding. Our temporal search mechanism captures event-level continuity by aggregating similarity scores across sequential video segments, enabling coherent retrieval of multi-event queries. Complementarily, a Google Image Search-based fallback module expands query representations with external web imagery, effectively bridging gaps in pretrained visual embeddings and improving robustness against out-of-distribution (OOD) queries. Together, these components advance the temporal reasoning and generalization capabilities of modern video retrieval systems, paving the way for more semantically aware and adaptive retrieval across large-scale video corpora.
Authors: Toan Le Ngo Thanh, Phat Ha Huu, Tan Nguyen Dang Duy, Thong Nguyen Le Minh, Anh Nguyen Nhu Tinh
Abstract: The exponential growth of video content has created an urgent need for efficient multimodal moment retrieval systems. However, existing approaches face three critical challenges: (1) fixed-weight fusion strategies fail across cross modal noise and ambiguous queries, (2) temporal modeling struggles to capture coherent event sequences while penalizing unrealistic gaps, and (3) systems require manual modality selection, reducing usability. We propose a unified multimodal moment retrieval system with three key innovations. First, a cascaded dual-embedding pipeline combines BEIT-3 and SigLIP for broad retrieval, refined by BLIP-2 based reranking to balance recall and precision. Second, a temporal-aware scoring mechanism applies exponential decay penalties to large temporal gaps via beam search, constructing coherent event sequences rather than isolated frames. Third, Agent-guided query decomposition (GPT-4o) automatically interprets ambiguous queries, decomposes them into modality specific sub-queries (visual/OCR/ASR), and performs adaptive score fusion eliminating manual modality selection. Qualitative analysis demonstrates that our system effectively handles ambiguous queries, retrieves temporally coherent sequences, and dynamically adapts fusion strategies, advancing interactive moment search capabilities.
Authors: Tiange Zhang, Xiandong Meng, Siwei Ma
Abstract: Recent advances in end-to-end video compression have shown promising results owing to their unified end-to-end learning optimization. However, such generalized frameworks often lack content-specific adaptation, leading to suboptimal compression performance. To address this, this paper proposes a content adaptive based motion alignment framework that improves performance by adapting encoding strategies to diverse content characteristics. Specifically, we first introduce a two-stage flow-guided deformable warping mechanism that refines motion compensation with coarse-to-fine offset prediction and mask modulation, enabling precise feature alignment. Second, we propose a multi-reference quality aware strategy that adjusts distortion weights based on reference quality, and applies it to hierarchical training to reduce error propagation. Third, we integrate a training-free module that downsamples frames by motion magnitude and resolution to obtain smooth motion estimation. Experimental results on standard test datasets demonstrate that our framework CAMA achieves significant improvements over state-of-the-art Neural Video Compression models, achieving a 24.95% BD-rate (PSNR) savings over our baseline model DCVC-TCM, while also outperforming reproduced DCVC-DC and traditional codec HM-16.25.
Authors: Siyuan Yao, Dongxiu Liu, Taotao Li, Shengjie Li, Wenqi Ren, Xiaochun Cao
Abstract: Building extraction from remote sensing images is a challenging task due to the complex structure variations of the buildings. Existing methods employ convolutional or self-attention blocks to capture the multi-scale features in the segmentation models, while the inherent gap of the feature pyramids and insufficient global-local feature integration leads to inaccurate, ambiguous extraction results. To address this issue, in this paper, we present an Uncertainty-Aggregated Global-Local Fusion Network (UAGLNet), which is capable to exploit high-quality global-local visual semantics under the guidance of uncertainty modeling. Specifically, we propose a novel cooperative encoder, which adopts hybrid CNN and transformer layers at different stages to capture the local and global visual semantics, respectively. An intermediate cooperative interaction block (CIB) is designed to narrow the gap between the local and global features when the network becomes deeper. Afterwards, we propose a Global-Local Fusion (GLF) module to complementarily fuse the global and local representations. Moreover, to mitigate the segmentation ambiguity in uncertain regions, we propose an Uncertainty-Aggregated Decoder (UAD) to explicitly estimate the pixel-wise uncertainty to enhance the segmentation accuracy. Extensive experiments demonstrate that our method achieves superior performance to other state-of-the-art methods. Our code is available at https://github.com/Dstate/UAGLNet
Authors: Luan Thanh Trinh, Kenji Doi, Atsuki Osanai
Abstract: Diffusion models have emerged as the leading approach for style transfer, yet they struggle with photo-realistic transfers, often producing painting-like results or missing detailed stylistic elements. Current methods inadequately address unwanted influence from original content styles and style reference content features. We introduce SCAdapter, a novel technique leveraging CLIP image space to effectively separate and integrate content and style features. Our key innovation systematically extracts pure content from content images and style elements from style references, ensuring authentic transfers. This approach is enhanced through three components: Controllable Style Adaptive Instance Normalization (CSAdaIN) for precise multi-style blending, KVS Injection for targeted style integration, and a style transfer consistency objective maintaining process coherence. Comprehensive experiments demonstrate SCAdapter significantly outperforms state-of-the-art methods in both conventional and diffusion-based baselines. By eliminating DDIM inversion and inference-stage optimization, our method achieves at least $2\times$ faster inference than other diffusion-based approaches, making it both more effective and efficient for practical applications.
Authors: Shengling Qin, Hao Yu, Chenxin Wu, Zheng Li, Yizhong Cao, Zhengyang Zhuge, Yuxin Zhou, Wentao Yao, Yi Zhang, Zhengheng Wang, Shuai Bai, Jianwei Zhang, Junyang Lin
Abstract: This paper presents VLCache, a cache reuse framework that exploits both Key-Value (KV) cache and encoder cache from prior multimodal inputs to eliminate costly recomputation when the same multimodal inputs recur. Unlike previous heuristic approaches, we formally identify the cumulative reuse error effect and demonstrate how to minimize the non-prefix cache reuse error effectively. We further analyze the varying importance of model layers and propose a dynamic, layer-aware recomputation strategy to balance accuracy and efficiency. Experimental results show that VLCache achieves an accuracy on par with full recomputation, while requiring only 2-5% of the tokens to compute, yielding 1.2x-16x TTFT speedups. The proposed VLCache pipeline has been integrated into SGLang, enabling significantly faster inference in practical deployments.
Authors: Ziheng Qin, Yuheng Ji, Renshuai Tao, Yuxuan Tian, Yuyang Liu, Yipu Wang, Xiaolong Zheng
Abstract: The pursuit of a universal AI-generated image (AIGI) detector often relies on aggregating data from numerous generators to improve generalization. However, this paper identifies a paradoxical phenomenon we term the Benefit then Conflict dilemma, where detector performance stagnates and eventually degrades as source diversity expands. Our systematic analysis, diagnoses this failure by identifying two core issues: severe data-level heterogeneity, which causes the feature distributions of real and synthetic images to increasingly overlap, and a critical model-level bottleneck from fixed, pretrained encoders that cannot adapt to the rising complexity. To address these challenges, we propose Generator-Aware Prototype Learning (GAPL), a framework that constrain representation with a structured learning paradigm. GAPL learns a compact set of canonical forgery prototypes to create a unified, low-variance feature space, effectively countering data heterogeneity.To resolve the model bottleneck, it employs a two-stage training scheme with Low-Rank Adaptation, enhancing its discriminative power while preserving valuable pretrained knowledge. This approach establishes a more robust and generalizable decision boundary. Through extensive experiments, we demonstrate that GAPL achieves state-of-the-art performance, showing superior detection accuracy across a wide variety of GAN and diffusion-based generators. Code is available at https://github.com/UltraCapture/GAPL
Authors: Wenjing lu, Zerui Tao, Dongping Zhang, Yuning Qiu, Yang Yang, Qibin Zhao
Abstract: CLIP delivers strong zero-shot classification but remains highly vulnerable to adversarial attacks. Previous work of adversarial fine-tuning largely focuses on matching the predicted logits between clean and adversarial examples, which overlooks uncertainty calibration and may degrade the zero-shot generalization. A common expectation in reliable uncertainty estimation is that predictive uncertainty should increase as inputs become more difficult or shift away from the training distribution. However, we frequently observe the opposite in the adversarial setting: perturbations not only degrade accuracy but also suppress uncertainty, leading to severe miscalibration and unreliable over-confidence. This overlooked phenomenon highlights a critical reliability gap beyond robustness. To bridge this gap, we propose a novel adversarial fine-tuning objective for CLIP considering both prediction accuracy and uncertainty alignments. By reparameterizing the output of CLIP as the concentration parameter of a Dirichlet distribution, we propose a unified representation that captures relative semantic structure and the magnitude of predictive confidence. Our objective aligns these distributions holistically under perturbations, moving beyond single-logit anchoring and restoring calibrated uncertainty. Experiments on multiple zero-shot classification benchmarks demonstrate that our approach effectively restores calibrated uncertainty and achieves competitive adversarial robustness while maintaining clean accuracy.
Authors: Yifan Pu, Yizeng Han, Zhiwei Tang, Jiasheng Tang, Fan Wang, Bohan Zhuang, Gao Huang
Abstract: Diffusion distillation has dramatically accelerated class-conditional image synthesis, but its applicability to open-ended text-to-image (T2I) generation is still unclear. We present the first systematic study that adapts and compares state-of-the-art distillation techniques on a strong T2I teacher model, FLUX.1-lite. By casting existing methods into a unified framework, we identify the key obstacles that arise when moving from discrete class labels to free-form language prompts. Beyond a thorough methodological analysis, we offer practical guidelines on input scaling, network architecture, and hyperparameters, accompanied by an open-source implementation and pretrained student models. Our findings establish a solid foundation for deploying fast, high-fidelity, and resource-efficient diffusion generators in real-world T2I applications. Code is available on github.com/alibaba-damo-academy/T2I-Distill.
Authors: Nikolai Goncharov, James L. Gray, Donald G. Dansereau
Abstract: Object tracking is an important step in robotics and reautonomous driving pipelines, which has to generalize to previously unseen and complex objects. Existing high-performing methods often rely on pre-captured object views to build explicit reference models, which restricts them to a fixed set of known objects. However, such reference models can struggle with visually complex appearance, reducing the quality of tracking. In this work, we introduce an object tracking method based on light field images that does not depend on a pre-trained model, while being robust to complex visual behavior, such as reflections. We extract semantic and geometric features from light field inputs using vision foundation models and convert them into view-dependent Gaussian splats. These splats serve as a unified object representation, supporting differentiable rendering and pose optimization. We further introduce a light field object tracking dataset containing challenging reflective objects with precise ground truth poses. Experiments demonstrate that our method is competitive with state-of-the-art model-based trackers in these difficult cases, paving the way toward universal object tracking in robotic systems. Code/data available at https://github.com/nagonch/LiFT-6DoF.
Authors: Xi Luo, Shixin Xu, Ying Xie, JianZhong Hu, Yuwei He, Yuhui Deng, Huaxiong Huang
Abstract: Accurate medical image analysis can greatly assist clinical diagnosis, but its effectiveness relies on high-quality expert annotations Obtaining pixel-level labels for medical images, particularly fundus images, remains costly and time-consuming. Meanwhile, despite the success of deep learning in medical imaging, the lack of interpretability limits its clinical adoption. To address these challenges, we propose TWLR, a two-stage framework for interpretable diabetic retinopathy (DR) assessment. In the first stage, a vision-language model integrates domain-specific ophthalmological knowledge into text embeddings to jointly perform DR grading and lesion classification, effectively linking semantic medical concepts with visual features. The second stage introduces an iterative severity regression framework based on weakly-supervised semantic segmentation. Lesion saliency maps generated through iterative refinement direct a progressive inpainting mechanism that systematically eliminates pathological features, effectively downgrading disease severity toward healthier fundus appearances. Critically, this severity regression approach achieves dual benefits: accurate lesion localization without pixel-level supervision and providing an interpretable visualization of disease-to-healthy transformations. Experimental results on the FGADR, DDR, and a private dataset demonstrate that TWLR achieves competitive performance in both DR classification and lesion segmentation, offering a more explainable and annotation-efficient solution for automated retinal image analysis.
Authors: Haoyu Wang, Lei Zhang, Wenrui Liu, Dengyang Jiang, Wei Wei, Chen Ding
Abstract: Given the inherently costly and time-intensive nature of pixel-level annotation, the generation of synthetic datasets comprising sufficiently diverse synthetic images paired with ground-truth pixel-level annotations has garnered increasing attention recently for training high-performance semantic segmentation models. However, existing methods necessitate to either predict pseudo annotations after image generation or generate images conditioned on manual annotation masks, which incurs image-annotation semantic inconsistency or scalability problem. To migrate both problems with one stone, we present a novel dataset generative diffusion framework for semantic segmentation, termed JoDiffusion. Firstly, given a standard latent diffusion model, JoDiffusion incorporates an independent annotation variational auto-encoder (VAE) network to map annotation masks into the latent space shared by images. Then, the diffusion model is tailored to capture the joint distribution of each image and its annotation mask conditioned on a text prompt. By doing these, JoDiffusion enables simultaneously generating paired images and semantically consistent annotation masks solely conditioned on text prompts, thereby demonstrating superior scalability. Additionally, a mask optimization strategy is developed to mitigate the annotation noise produced during generation. Experiments on Pascal VOC, COCO, and ADE20K datasets show that the annotated dataset generated by JoDiffusion yields substantial performance improvements in semantic segmentation compared to existing methods.
Authors: Xinjie Li, Zhimin Chen, Rui Zhao, Florian Schiffers, Zhenyu Liao, Vimal Bhat
Abstract: Recent unified models for joint understanding and generation have significantly advanced visual generation capabilities. However, their focus on conventional tasks like text-to-video generation has left the temporal reasoning potential of unified models largely underexplored. To address this gap, we introduce Next Scene Prediction (NSP), a new task that pushes unified video models toward temporal and causal reasoning. Unlike text-to-video generation, NSP requires predicting plausible futures from preceding context, demanding deeper understanding and reasoning. To tackle this task, we propose a unified framework combining Qwen-VL for comprehension and LTX for synthesis, bridged by a latent query embedding and a connector module. This model is trained in three stages on our newly curated, large-scale NSP dataset: text-to-video pre-training, supervised fine-tuning, and reinforcement learning (via GRPO) with our proposed causal consistency reward. Experiments demonstrate our model achieves state-of-the-art performance on our benchmark, advancing the capability of generalist multimodal systems to anticipate what happens next.
Authors: Tomoya Tanaka, Tomonori Ikeda, Ryo Yonemoto
Abstract: This study presents the first comprehensive evaluation of spatial generalization techniques, which are essential for the practical deployment of deep learning-based radio-frequency (RF) sensing. Focusing on people counting in indoor environments using frequency-modulated continuous-wave (FMCW) multiple-input multiple-output (MIMO) radar, we systematically investigate a broad set of approaches, including amplitude-based statistical preprocessing (sigmoid weighting and threshold zeroing), frequency-domain filtering, autoencoder-based background suppression, data augmentation strategies, and transfer learning. Experimental results collected across two environments with different layouts demonstrate that sigmoid-based amplitude weighting consistently achieves superior cross-environment performance, yielding 50.1% and 55.2% reductions in root-mean-square error (RMSE) and mean absolute error (MAE), respectively, compared with baseline methods. Data augmentation provides additional though modest benefits, with improvements up to 8.8% in MAE. By contrast, transfer learning proves indispensable for large spatial shifts, achieving 82.1% and 91.3% reductions in RMSE and MAE, respectively, with 540 target-domain samples. Taken together, these findings establish a highly practical direction for developing radar sensing systems capable of maintaining robust accuracy under spatial variations by integrating deep learning models with amplitude-based preprocessing and efficient transfer learning.
Authors: Cheeun Hong, German Barquero, Fadime Sener, Markos Georgopoulos, Edgar Sch\"onfeld, Stefan Popov, Yuming Du, Oscar Ma\~nas, Albert Pumarola
Abstract: Instructional video generation is an emerging task that aims to synthesize coherent demonstrations of procedural activities from textual descriptions. Such capability has broad implications for content creation, education, and human-AI interaction, yet existing video diffusion models struggle to maintain temporal consistency and controllability across long sequences of multiple action steps. We introduce a pipeline for future-driven streaming instructional video generation, dubbed SneakPeek, a diffusion-based autoregressive framework designed to generate precise, stepwise instructional videos conditioned on an initial image and structured textual prompts. Our approach introduces three key innovations to enhance consistency and controllability: (1) predictive causal adaptation, where a causal model learns to perform next-frame prediction and anticipate future keyframes; (2) future-guided self-forcing with a dual-region KV caching scheme to address the exposure bias issue at inference time; (3) multi-prompt conditioning, which provides fine-grained and procedural control over multi-step instructions. Together, these components mitigate temporal drift, preserve motion consistency, and enable interactive video generation where future prompt updates dynamically influence ongoing streaming video generation. Experimental results demonstrate that our method produces temporally coherent and semantically faithful instructional videos that accurately follow complex, multi-step task descriptions.
Authors: Hongzhe Bi, Hengkai Tan, Shenghao Xie, Zeyuan Wang, Shuhe Huang, Haitian Liu, Ruowen Zhao, Yao Feng, Chendong Xiang, Yinze Rong, Hongyan Zhao, Hanyu Liu, Zhizhong Su, Lei Ma, Hang Su, Jun Zhu
Abstract: While a general embodied agent must function as a unified system, current methods are built on isolated models for understanding, world modeling, and control. This fragmentation prevents unifying multimodal generative capabilities and hinders learning from large-scale, heterogeneous data. In this paper, we propose Motus, a unified latent action world model that leverages existing general pretrained models and rich, sharable motion information. Motus introduces a Mixture-of-Transformer (MoT) architecture to integrate three experts (i.e., understanding, video generation, and action) and adopts a UniDiffuser-style scheduler to enable flexible switching between different modeling modes (i.e., world models, vision-language-action models, inverse dynamics models, video generation models, and video-action joint prediction models). Motus further leverages the optical flow to learn latent actions and adopts a recipe with three-phase training pipeline and six-layer data pyramid, thereby extracting pixel-level "delta action" and enabling large-scale action pretraining. Experiments show that Motus achieves superior performance against state-of-the-art methods in both simulation (a +15% improvement over X-VLA and a +45% improvement over Pi0.5) and real-world scenarios(improved by +11~48%), demonstrating unified modeling of all functionalities and priors significantly benefits downstream robotic tasks.
Authors: Tomoya Tanaka, Tomonori Ikeda, Ryo Yonemoto
Abstract: This study presents the first comprehensive comparison of rule-based methods, traditional machine learning models, and deep learning models in radio wave sensing with frequency modulated continuous wave multiple input multiple output radar. We systematically evaluated five approaches in two indoor environments with distinct layouts: a rule-based connected component method; three traditional machine learning models, namely k-nearest neighbors, random forest, and support vector machine; and a deep learning model combining a convolutional neural network and long short term memory. In the training environment, the convolutional neural network long short term memory model achieved the highest accuracy, while traditional machine learning models provided moderate performance. In a new layout, however, all learning based methods showed significant degradation, whereas the rule-based method remained stable. Notably, for binary detection of presence versus absence of people, all models consistently achieved high accuracy across layouts. These results demonstrate that high capacity models can produce fine grained outputs with high accuracy in the same environment, but they are vulnerable to domain shift. In contrast, rule-based methods cannot provide fine grained outputs but exhibit robustness against domain shift. Moreover, regardless of the model type, a clear trade off was revealed between spatial generalization performance and output granularity.
Authors: Hao Chen, Yiwei Wang, Songze Li
Abstract: Concept erasure, which fine-tunes diffusion models to remove undesired or harmful visual concepts, has become a mainstream approach to mitigating unsafe or illegal image generation in text-to-image models.However, existing removal methods typically adopt a unidirectional erasure strategy by either suppressing the target concept or reinforcing safe alternatives, making it difficult to achieve a balanced trade-off between concept removal and generation quality. To address this limitation, we propose a novel Bidirectional Image-Guided Concept Erasure (Bi-Erasing) framework that performs concept suppression and safety enhancement simultaneously. Specifically, based on the joint representation of text prompts and corresponding images, Bi-Erasing introduces two decoupled image branches: a negative branch responsible for suppressing harmful semantics and a positive branch providing visual guidance for safe alternatives. By jointly optimizing these complementary directions, our approach achieves a balance between erasure efficacy and generation usability. In addition, we apply mask-based filtering to the image branches to prevent interference from irrelevant content during the erasure process. Across extensive experiment evaluations, the proposed Bi-Erasing outperforms baseline methods in balancing concept removal effectiveness and visual fidelity.
Authors: Tong Wei, Yijun Yang, Changhao Zhang, Junliang Xing, Yuanchun Shi, Zongqing Lu, Deheng Ye
Abstract: Multi-turn reinforcement learning (RL) for multi-modal agents built upon vision-language models (VLMs) is hampered by sparse rewards and long-horizon credit assignment. Recent methods densify the reward by querying a teacher that provides step-level feedback, e.g., Guided Thought Reinforcement (GTR) and On-Policy Distillation, but rely on costly, often privileged models as the teacher, limiting practicality and reproducibility. We introduce GTR-Turbo, a highly efficient upgrade to GTR, which matches the performance without training or querying an expensive teacher model. Specifically, GTR-Turbo merges the weights of checkpoints produced during the ongoing RL training, and then uses this merged model as a "free" teacher to guide the subsequent RL via supervised fine-tuning or soft logit distillation. This design removes dependence on privileged VLMs (e.g., GPT or Gemini), mitigates the "entropy collapse" observed in prior work, and keeps training stable. Across diverse visual agentic tasks, GTR-Turbo improves the accuracy of the baseline model by 10-30% while reducing wall-clock training time by 50% and compute cost by 60% relative to GTR.
Authors: Jaeyoon Kim, Yoonki Cho, Sung-Eui Yoon
Abstract: Visual Place Recognition (VPR) has advanced significantly with high-capacity foundation models like DINOv2, achieving remarkable performance. Nonetheless, their substantial computational cost makes deployment on resource-constrained devices impractical. In this paper, we introduce an efficient asymmetric VPR framework that incorporates a high-capacity gallery model for offline feature extraction with a lightweight query network for online processing. A key challenge in this setting is ensuring compatibility between these heterogeneous networks, which conventional approaches address through computationally expensive k-NN-based compatible training. To overcome this, we propose a geographical memory bank that structures gallery features using geolocation metadata inherent in VPR databases, eliminating the need for exhaustive k-NN computations. Additionally, we introduce an implicit embedding augmentation technique that enhances the query network to model feature variations despite its limited capacity. Extensive experiments demonstrate that our method not only significantly reduces computational costs but also outperforms existing asymmetric retrieval techniques, establishing a new aspect for VPR in resource-limited environments. The code is available at https://github.com/jaeyoon1603/AsymVPR
Authors: Zizhi Chen, Yizhen Gao, Minghao Han, Yizhou Liu, Zhaoyu Chen, Dingkang Yang, Lihua Zhang
Abstract: Multimodal biomedical Vision-Language Models (VLMs) exhibit immense potential in the field of Continual Learning (CL). However, they confront a core dilemma: how to preserve fine-grained intra-modality features while bridging the significant domain gap across different modalities. To address this challenge, we propose a comprehensive framework. Leveraging our 18-million multimodal and comprehensive medical retrieval database derived from PubMed scientific papers, we pioneer the integration of Retrieval-Augmented Generation (RAG) into CL. Specifically, we employ a multi-modal, multi-layer RAG system that provides real-time guidance for model fine-tuning through dynamic, on-demand knowledge retrieval. Building upon this, we introduce a dynamic knowledge distillation framework. This framework precisely resolves the aforementioned core dilemma by dynamically modulating the importance of the parameter space, the granularity of the distilled knowledge, and the data distribution of the reference dataset in accordance with the required level of detail. To thoroughly validate the clinical value of our strategy, we have designed a more rigorous \textbf{M}edical Generalist Task Incremental Learning (MGTIL) benchmark. This benchmark is engineered to simultaneously evaluate the model's capacity for adaptation to significant domain shifts, retention of subtle intra-domain features, and real-time learning of novel and complex medical tasks. Extensive experimental results demonstrate that our proposed method achieves state-of-the-art (SOTA) performance across all metrics. The code is provided in the supplementary materials.
Authors: Mohaiminul Islam Bhuiyan, Chan Hue Wah, Nur Shazwani Kamarudin, Nur Hafieza Ismail, Ahmad Fakhri Ab Nasir
Abstract: This study provides an overview of heart disease prediction using an intelligent system. Predicting disease accurately is crucial in the medical field, but traditional methods relying solely on a doctor's experience often lack precision. To address this limitation, intelligent systems are applied as an alternative to traditional approaches. While various intelligent system methods exist, this study focuses on three: Fuzzy Logic, Neural Networks, and Case-Based Reasoning (CBR). A comparison of these techniques in terms of accuracy was conducted, and ultimately, Case-Based Reasoning (CBR) was selected for heart disease prediction. In the prediction phase, the heart disease dataset underwent data pre-processing to clean the data and data splitting to separate it into training and testing sets. The chosen intelligent system was then employed to predict heart disease outcomes based on the processed data. The experiment concluded with Case-Based Reasoning (CBR) achieving a notable accuracy rate of 97.95% in predicting heart disease. The findings also revealed that the probability of heart disease was 57.76% for males and 42.24% for females. Further analysis from related studies suggests that factors such as smoking and alcohol consumption are significant contributors to heart disease, particularly among males.
Authors: Saumyaranjan Mohanty, Aravind Reddy, Konda Reddy Mopuri
Abstract: In Dataset Condensation, the goal is to synthesize a small dataset that replicates the training utility of a large original dataset. Existing condensation methods synthesize datasets with significant redundancy, so there is a dire need to reduce redundancy and improve the diversity of the synthesized datasets. To tackle this, we propose an intuitive Diversity Regularizer (DiRe) composed of cosine similarity and Euclidean distance, which can be applied off-the-shelf to various state-of-the-art condensation methods. Through extensive experiments, we demonstrate that the addition of our regularizer improves state-of-the-art condensation methods on various benchmark datasets from CIFAR-10 to ImageNet-1K with respect to generalization and diversity metrics.
Authors: Ziqiang Zhu, Bowei Yang
Abstract: Change detection (CD) identifies scene changes from multi-temporal observations and is widely used in urban development and environmental monitoring. Most existing CD methods rely on supervised learning, making performance strongly dataset-dependent and incurring high annotation costs; they typically focus on a few predefined categories and generalize poorly to diverse scenes. With the rise of vision foundation models such as SAM2 and CLIP, new opportunities have emerged to relax these constraints. We propose Unified Open-Vocabulary Change Detection (UniVCD), an unsupervised, open-vocabulary change detection method built on frozen SAM2 and CLIP. UniVCD detects category-agnostic changes across diverse scenes and imaging geometries without any labeled data or paired change images. A lightweight feature alignment module is introduced to bridge the spatially detailed representations from SAM2 and the semantic priors from CLIP, enabling high-resolution, semantically aware change estimation while keeping the number of trainable parameters small. On top of this, a streamlined post-processing pipeline is further introduced to suppress noise and pseudo-changes, improving the detection accuracy for objects with well-defined boundaries. Experiments on several public BCD (Binary Change Detection) and SCD (Semantic Change Detection) benchmarks show that UniVCD achieves consistently strong performance and matches or surpasses existing open-vocabulary CD methods in key metrics such as F1 and IoU. The results demonstrate that unsupervised change detection with frozen vision foundation models and lightweight multi-modal alignment is a practical and effective paradigm for open-vocabulary CD. Code and pretrained models will be released at https://github.com/Die-Xie/UniVCD.
Authors: Feng Zhang, Zezhong Tan, Xinhong Ma, Ziqiang Dong, Xi Leng, Jianfei Zhao, Xin Sun, Yang Yang
Abstract: To combine the advantages of Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), recent methods have integrated ''hints'' into post-training, which are prefix segments of complete reasoning trajectories, aiming for powerful knowledge expansion and reasoning generalization. However, existing hint-based RL methods typically ignore difficulty when scheduling hint ratios and estimating relative advantages, leading to unstable learning and excessive imitation of off-policy hints. In this work, we propose ADHint, which treats difficulty as a key factor in both hint-ratio schedule and relative-advantage estimation to achieve a better trade-off between exploration and imitation. Specifically, we propose Adaptive Hint with Sample Difficulty Prior, which evaluates each sample's difficulty under the policy model and accordingly schedules an appropriate hint ratio to guide its rollouts. We also introduce Consistency-based Gradient Modulation and Selective Masking for Hint Preservation to modulate token-level gradients within hints, preventing biased and destructive updates. Additionally, we propose Advantage Estimation with Rollout Difficulty Posterior, which leverages the relative difficulty of rollouts with and without hints to estimate their respective advantages, thereby achieving more balanced updates. Extensive experiments across diverse modalities, model scales, and domains demonstrate that ADHint delivers superior reasoning ability and out-of-distribution generalization, consistently surpassing existing methods in both pass@1 and avg@8. Our code and dataset will be made publicly available upon paper acceptance.
Authors: Wenjing Lu, Yi Hong, Yang Yang
Abstract: Vision foundation models have demonstrated strong generalization in medical image segmentation by leveraging large-scale, heterogeneous pretraining. However, they often struggle to generalize to specialized clinical tasks under limited annotations or rare pathological variations, due to a mismatch between general priors and task-specific requirements. To address this, we propose Uncertainty-informed Collaborative Learning (UnCoL), a dual-teacher framework that harmonizes generalization and specialization in semi-supervised medical image segmentation. Specifically, UnCoL distills both visual and semantic representations from a frozen foundation model to transfer general knowledge, while concurrently maintaining a progressively adapting teacher to capture fine-grained and task-specific representations. To balance guidance from both teachers, pseudo-label learning in UnCoL is adaptively regulated by predictive uncertainty, which selectively suppresses unreliable supervision and stabilizes learning in ambiguous regions. Experiments on diverse 2D and 3D segmentation benchmarks show that UnCoL consistently outperforms state-of-the-art semi-supervised methods and foundation model baselines. Moreover, our model delivers near fully supervised performance with markedly reduced annotation requirements.
Authors: Yan Zhang, Baoxin Li, Han Sun, Yuhang Gao, Mingtai Zhang, Pei Wang
Abstract: Forest pests threaten ecosystem stability, requiring efficient monitoring. To overcome the limitations of traditional methods in large-scale, fine-grained detection, this study focuses on accurately identifying infected trees and analyzing infestation patterns. We propose FID-Net, a deep learning model that detects pest-affected trees from UAV visible-light imagery and enables infestation analysis via three spatial metrics. Based on YOLOv8n, FID-Net introduces a lightweight Feature Enhancement Module (FEM) to extract disease-sensitive cues, an Adaptive Multi-scale Feature Fusion Module (AMFM) to align and fuse dual-branch features (RGB and FEM-enhanced), and an Efficient Channel Attention (ECA) mechanism to enhance discriminative information efficiently. From detection results, we construct a pest situation analysis framework using: (1) Kernel Density Estimation to locate infection hotspots; (2) neighborhood evaluation to assess healthy trees' infection risk; (3) DBSCAN clustering to identify high-density healthy clusters as priority protection zones. Experiments on UAV imagery from 32 forest plots in eastern Tianshan, China, show that FID-Net achieves 86.10% precision, 75.44% recall, 82.29% mAP@0.5, and 64.30% mAP@0.5:0.95, outperforming mainstream YOLO models. Analysis confirms infected trees exhibit clear clustering, supporting targeted forest protection. FID-Net enables accurate tree health discrimination and, combined with spatial metrics, provides reliable data for intelligent pest monitoring, early warning, and precise management.
Authors: Zhijian He, Feifei Liu, Yuwei Li, Zhanpeng Liu, Jintao Cheng, Xieyuanli Chen, Xiaoyu Tang
Abstract: Multi-modal 3D object detection is important for reliable perception in robotics and autonomous driving. However, its effectiveness remains limited under adverse weather conditions due to weather-induced distortions and misalignment between different data modalities. In this work, we propose DiffFusion, a novel framework designed to enhance robustness in challenging weather through diffusion-based restoration and adaptive cross-modal fusion. Our key insight is that diffusion models possess strong capabilities for denoising and generating data that can adapt to various weather conditions. Building on this, DiffFusion introduces Diffusion-IR restoring images degraded by weather effects and Point Cloud Restoration (PCR) compensating for corrupted LiDAR data using image object cues. To tackle misalignments between two modalities, we develop Bidirectional Adaptive Fusion and Alignment Module (BAFAM). It enables dynamic multi-modal fusion and bidirectional bird's-eye view (BEV) alignment to maintain consistent spatial correspondence. Extensive experiments on three public datasets show that DiffFusion achieves state-of-the-art robustness under adverse weather while preserving strong clean-data performance. Zero-shot results on the real-world DENSE dataset further validate its generalization. The implementation of our DiffFusion will be released as open-source.
Authors: Vivek Alumootil, Tuan-Anh Vu, M. Khalid Jawed
Abstract: Current methods for dense 3D point tracking in dynamic scenes typically rely on pairwise processing, require known camera poses, or assume a temporal ordering to input frames, constraining their flexibility and applicability. Additionally, recent advances have successfully enabled efficient 3D reconstruction from large-scale, unposed image collections, underscoring opportunities for unified approaches to dynamic scene understanding. Motivated by this, we propose DePT3R, a novel framework that simultaneously performs dense point tracking and 3D reconstruction of dynamic scenes from multiple images in a single forward pass. This multi-task learning is achieved by extracting deep spatio-temporal features with a powerful backbone and regressing pixel-wise maps with dense prediction heads. Crucially, DePT3R operates without requiring camera poses, substantially enhancing its adaptability and efficiency-especially important in dynamic environments with rapid changes. We validate DePT3R on several challenging benchmarks involving dynamic scenes, demonstrating strong performance and significant improvements in memory efficiency over existing state-of-the-art methods. Data and codes are available via the open repository: https://github.com/StructuresComp/DePT3R
Authors: Shanghua Liu, Majharulislam Babor, Christoph Verduyn, Breght Vandenberghe, Bruno Betoni Parodi, Cornelia Weltzien, Marina M. -C. H\"ohne
Abstract: High resolution phenotyping at the level of individual leaves offers fine-grained insights into plant development and stress responses. However, the full potential of accurate leaf tracking over time remains largely unexplored due to the absence of robust tracking methods-particularly for structurally complex crops such as canola. Existing plant-specific tracking methods are typically limited to small-scale species or rely on constrained imaging conditions. In contrast, generic multi-object tracking (MOT) methods are not designed for dynamic biological scenes. Progress in the development of accurate leaf tracking models has also been hindered by a lack of large-scale datasets captured under realistic conditions. In this work, we introduce CanolaTrack, a new benchmark dataset comprising 5,704 RGB images with 31,840 annotated leaf instances spanning the early growth stages of 184 canola plants. To enable accurate leaf tracking over time, we introduce LeafTrackNet, an efficient framework that combines a YOLOv10-based leaf detector with a MobileNetV3-based embedding network. During inference, leaf identities are maintained over time through an embedding-based memory association strategy. LeafTrackNet outperforms both plant-specific trackers and state-of-the-art MOT baselines, achieving a 9% HOTA improvement on CanolaTrack. With our work we provide a new standard for leaf-level tracking under realistic conditions and we provide CanolaTrack - the largest dataset for leaf tracking in agriculture crops, which will contribute to future research in plant phenotyping. Our code and dataset are publicly available at https://github.com/shl-shawn/LeafTrackNet.
Authors: Chun Kit Wong, Paraskevas Pegios, Nina Weng, Emilie Pi Fogtmann Sejer, Martin Gr{\o}nneb{\ae}k Tolsgaard, Anders Nymark Christensen, Aasa Feragen
Abstract: Deep learning models in medical imaging are susceptible to shortcut learning, relying on confounding metadata (e.g., scanner model) that is often encoded in image embeddings. The crucial question is whether the model actively utilizes this encoded information for its final prediction. We introduce Weight Space Correlation Analysis, an interpretable methodology that quantifies feature utilization by measuring the alignment between the classification heads of a primary clinical task and auxiliary metadata tasks. We first validate our method by successfully detecting artificially induced shortcut learning. We then apply it to probe the feature utilization of an SA-SonoNet model trained for Spontaneous Preterm Birth (sPTB) prediction. Our analysis confirmed that while the embeddings contain substantial metadata, the sPTB classifier's weight vectors were highly correlated with clinically relevant factors (e.g., birth weight) but decoupled from clinically irrelevant acquisition factors (e.g. scanner). Our methodology provides a tool to verify model trustworthiness, demonstrating that, in the absence of induced bias, the clinical model selectively utilizes features related to the genuine clinical signal.
Authors: Sangmin Hong, Suyoung Lee, Kyoung Mu Lee
Abstract: The problem of depth completion involves predicting a dense depth image from a single sparse depth map and an RGB image. Unsupervised depth completion methods have been proposed for various datasets where ground truth depth data is unavailable and supervised methods cannot be applied. However, these models require auxiliary data to estimate depth values, which is far from real scenarios. Monocular depth estimation (MDE) models can produce a plausible relative depth map from a single image, but there is no work to properly combine the sparse depth map with MDE for depth completion; a simple affine transformation to the depth map will yield a high error since MDE are inaccurate at estimating depth difference between objects. We introduce StarryGazer, a domain-agnostic framework that predicts dense depth images from a single sparse depth image and an RGB image without relying on ground-truth depth by leveraging the power of large MDE models. First, we employ a pre-trained MDE model to produce relative depth images. These images are segmented and randomly rescaled to form synthetic pairs for dense pseudo-ground truth and corresponding sparse depths. A refinement network is trained with the synthetic pairs, incorporating the relative depth maps and RGB images to improve the model's accuracy and robustness. StarryGazer shows superior results over existing unsupervised methods and transformed MDE results on various datasets, demonstrating that our framework exploits the power of MDE models while appropriately fixing errors using sparse depth information.
Authors: Peter Kocsis (Technical University of Munich), Lukas H\"ollein (Technical University of Munich), Matthias Nie{\ss}ner (Technical University of Munich)
Abstract: We introduce Intrinsic Image Fusion, a method that reconstructs high-quality physically based materials from multi-view images. Material reconstruction is highly underconstrained and typically relies on analysis-by-synthesis, which requires expensive and noisy path tracing. To better constrain the optimization, we incorporate single-view priors into the reconstruction process. We leverage a diffusion-based material estimator that produces multiple, but often inconsistent, candidate decompositions per view. To reduce the inconsistency, we fit an explicit low-dimensional parametric function to the predictions. We then propose a robust optimization framework using soft per-view prediction selection together with confidence-based soft multi-view inlier set to fuse the most consistent predictions of the most confident views into a consistent parametric material space. Finally, we use inverse path tracing to optimize for the low-dimensional parameters. Our results outperform state-of-the-art methods in material disentanglement on both synthetic and real scenes, producing sharp and clean reconstructions suitable for high-quality relighting.
Authors: Xianchao Guan, Zhiyuan Fan, Yifeng Wang, Fuqiang Chen, Yanjiang Zhou, Zengyang Che, Hongxue Meng, Xin Li, Yaowei Wang, Hongpeng Wang, Min Zhang, Heng Tao Shen, Zheng Zhang, Yongbing Zhang
Abstract: The development of clinical-grade artificial intelligence in pathology is limited by the scarcity of diverse, high-quality annotated datasets. Generative models offer a potential solution but suffer from semantic instability and morphological hallucinations that compromise diagnostic reliability. To address this challenge, we introduce a Correlation-Regulated Alignment Framework for Tissue Synthesis (CRAFTS), the first generative foundation model for pathology-specific text-to-image synthesis. By leveraging a dual-stage training strategy on approximately 2.8 million image-caption pairs, CRAFTS incorporates a novel alignment mechanism that suppresses semantic drift to ensure biological accuracy. This model generates diverse pathological images spanning 30 cancer types, with quality rigorously validated by objective metrics and pathologist evaluations. Furthermore, CRAFTS-augmented datasets enhance the performance across various clinical tasks, including classification, cross-modal retrieval, self-supervised learning, and visual question answering. In addition, coupling CRAFTS with ControlNet enables precise control over tissue architecture from inputs such as nuclear segmentation masks and fluorescence images. By overcoming the critical barriers of data scarcity and privacy concerns, CRAFTS provides a limitless source of diverse, annotated histology data, effectively unlocking the creation of robust diagnostic tools for rare and complex cancer phenotypes.
Authors: Hongxuan Sun, Tao Wu
Abstract: Semantic segmentation requires a holistic understanding of the physical world, as it assigns semantic labels to spatially continuous and structurally coherent objects rather than to isolated pixels. However, existing data-free knowledge distillation (DFKD) methods-primarily designed for classification-often disregard this continuity, resulting in significant performance degradation when applied directly to segmentation tasks. In this paper, we introduce DFSS, a novel data-free distillation framework tailored for semantic segmentation. Unlike prior approaches that treat pixels independently, DFSS respects the structural and contextual continuity of real-world scenes. Our key insight is to leverage Batch Normalization (BN) statistics from a teacher model to guide Approximate Distribution Sampling (ADS), enabling the selection of data that better reflects the original training distribution-without relying on potentially misleading teacher predictions. Additionally, we propose Weighted Distribution Progressive Distillation (WDPD), which dynamically prioritizes reliable samples that are more closely aligned with the original data distribution early in training and gradually incorporates more challenging cases, mirroring the natural progression of learning in human perception. Extensive experiments on standard benchmarks demonstrate that DFSS consistently outperforms existing data-free distillation methods for semantic segmentation, achieving state-of-the-art results with significantly reduced reliance on auxiliary data.
Authors: Minghui Hou, Wei-Hsing Huang, Shaofeng Liang, Daizong Liu, Tai-Hao Wen, Gang Wang, Runwei Guan, Weiping Ding
Abstract: Vision-language models enable the understanding and reasoning of complex traffic scenarios through multi-source information fusion, establishing it as a core technology for autonomous driving. However, existing vision-language models are constrained by the image understanding paradigm in 2D plane, which restricts their capability to perceive 3D spatial information and perform deep semantic fusion, resulting in suboptimal performance in complex autonomous driving environments. This study proposes MMDrive, an multimodal vision-language model framework that extends traditional image understanding to a generalized 3D scene understanding framework. MMDrive incorporates three complementary modalities, including occupancy maps, LiDAR point clouds, and textual scene descriptions. To this end, it introduces two novel components for adaptive cross-modal fusion and key information extraction. Specifically, the Text-oriented Multimodal Modulator dynamically weights the contributions of each modality based on the semantic cues in the question, guiding context-aware feature integration. The Cross-Modal Abstractor employs learnable abstract tokens to generate compact, cross-modal summaries that highlight key regions and essential semantics. Comprehensive evaluations on the DriveLM and NuScenes-QA benchmarks demonstrate that MMDrive achieves significant performance gains over existing vision-language models for autonomous driving, with a BLEU-4 score of 54.56 and METEOR of 41.78 on DriveLM, and an accuracy score of 62.7% on NuScenes-QA. MMDrive effectively breaks the traditional image-only understanding barrier, enabling robust multimodal reasoning in complex driving environments and providing a new foundation for interpretable autonomous driving scene understanding.
Authors: Gong Chen, Chaokun Zhang, Pengcheng Lv, Xiaohui Xie
Abstract: Collaborative perception has garnered significant attention as a crucial technology to overcome the perceptual limitations of single-agent systems. Many state-of-the-art (SOTA) methods have achieved communication efficiency and high performance via intermediate fusion. However, they share a critical vulnerability: their performance degrades under adverse communication conditions due to the misalignment induced by data transmission, which severely hampers their practical deployment. To bridge this gap, we re-examine different fusion paradigms, and recover that the strengths of intermediate and late fusion are not a trade-off, but a complementary pairing. Based on this key insight, we propose CoRA, a novel collaborative robust architecture with a hybrid approach to decouple performance from robustness with low communication. It is composed of two components: a feature-level fusion branch and an object-level correction branch. Its first branch selects critical features and fuses them efficiently to ensure both performance and scalability. The second branch leverages semantic relevance to correct spatial displacements, guaranteeing resilience against pose errors. Experiments demonstrate the superiority of CoRA. Under extreme scenarios, CoRA improves upon its baseline performance by approximately 19% in AP@0.7 with more than 5x less communication volume, which makes it a promising solution for robust collaborative perception.
Authors: Zhuo Chen, Chengqun Yang, Zhuo Su, Zheng Lv, Jingnan Gao, Xiaoyuan Zhang, Xiaokang Yang, Yichao Yan
Abstract: Face relighting aims to synthesize realistic portraits under novel illumination while preserving identity and geometry. However, progress remains constrained by the limited availability of large-scale, physically consistent illumination data. To address this, we introduce POLAR, a large-scale and physically calibrated One-Light-at-a-Time (OLAT) dataset containing over 200 subjects captured under 156 lighting directions, multiple views, and diverse expressions. Building upon POLAR, we develop a flow-based generative model POLARNet that predicts per-light OLAT responses from a single portrait, capturing fine-grained and direction-aware illumination effects while preserving facial identity. Unlike diffusion or background-conditioned methods that rely on statistical or contextual cues, our formulation models illumination as a continuous, physically interpretable transformation between lighting states, enabling scalable and controllable relighting. Together, POLAR and POLARNet form a unified illumination learning framework that links real data, generative synthesis, and physically grounded relighting, establishing a self-sustaining "chicken-and-egg" cycle for scalable and reproducible portrait illumination.
Authors: Francesco Ragusa, Michele Mazzamuto, Rosario Forte, Irene D'Ambra, James Fort, Jakob Engel, Antonino Furnari, Giovanni Maria Farinella
Abstract: We present Ego-EXTRA, a video-language Egocentric Dataset for EXpert-TRAinee assistance. Ego-EXTRA features 50 hours of unscripted egocentric videos of subjects performing procedural activities (the trainees) while guided by real-world experts who provide guidance and answer specific questions using natural language. Following a ``Wizard of OZ'' data collection paradigm, the expert enacts a wearable intelligent assistant, looking at the activities performed by the trainee exclusively from their egocentric point of view, answering questions when asked by the trainee, or proactively interacting with suggestions during the procedures. This unique data collection protocol enables Ego-EXTRA to capture a high-quality dialogue in which expert-level feedback is provided to the trainee. Two-way dialogues between experts and trainees are recorded, transcribed, and used to create a novel benchmark comprising more than 15k high-quality Visual Question Answer sets, which we use to evaluate Multimodal Large Language Models. The results show that Ego-EXTRA is challenging and highlight the limitations of current models when used to provide expert-level assistance to the user. The Ego-EXTRA dataset is publicly available to support the benchmark of egocentric video-language assistants: https://fpv-iplab.github.io/Ego-EXTRA/.
Authors: Foivos Paraperas Papantoniou, Stathis Galanakis, Rolandos Alexandros Potamias, Bernhard Kainz, Stefanos Zafeiriou
Abstract: This paper presents STARCaster, an identity-aware spatio-temporal video diffusion model that addresses both speech-driven portrait animation and free-viewpoint talking portrait synthesis, given an identity embedding or reference image, within a unified framework. Existing 2D speech-to-video diffusion models depend heavily on reference guidance, leading to limited motion diversity. At the same time, 3D-aware animation typically relies on inversion through pre-trained tri-plane generators, which often leads to imperfect reconstructions and identity drift. We rethink reference- and geometry-based paradigms in two ways. First, we deviate from strict reference conditioning at pre-training by introducing softer identity constraints. Second, we address 3D awareness implicitly within the 2D video domain by leveraging the inherent multi-view nature of video data. STARCaster adopts a compositional approach progressing from ID-aware motion modeling, to audio-visual synchronization via lip reading-based supervision, and finally to novel view animation through temporal-to-spatial adaptation. To overcome the scarcity of 4D audio-visual data, we propose a decoupled learning approach in which view consistency and temporal coherence are trained independently. A self-forcing training scheme enables the model to learn from longer temporal contexts than those generated at inference, mitigating the overly static animations common in existing autoregressive approaches. Comprehensive evaluations demonstrate that STARCaster generalizes effectively across tasks and identities, consistently surpassing prior approaches in different benchmarks.
Authors: Juil Koo, Daehyeon Choi, Sangwoo Youn, Phillip Y. Lee, Minhyuk Sung
Abstract: Vision Language Models (VLMs) excel at visual question answering (VQA) but remain limited to snapshot vision, reasoning from static images. In contrast, embodied agents require ambulatory vision, actively moving to obtain more informative views. We introduce Visually Grounded Active View Selection (VG-AVS), a task that selects the most informative next viewpoint using only the visual information in the current image, without relying on scene memory or external knowledge. To support this task, we construct a synthetic dataset with automatically generated paired query-target views and question-answer prompts. We also propose a framework that fine-tunes pretrained VLMs through supervised fine-tuning (SFT) followed by RL-based policy optimization. Our approach achieves strong question answering performance based on viewpoint selection and generalizes robustly to unseen synthetic and real scenes. Furthermore, incorporating our learned VG-AVS framework into existing scene-exploration-based EQA systems improves downstream question-answering accuracy.
Authors: Yan Li, Lin Liu, Xiaopeng Zhang, Wei Xue, Wenhan Luo, Yike Guo, Qi Tian
Abstract: Instruction-based image editing with diffusion models has achieved impressive results, yet existing methods struggle with fine-grained instructions specifying precise attributes such as colors, positions, and quantities. While recent approaches employ Group Relative Policy Optimization (GRPO) for alignment, they optimize only at individual sampling steps, providing sparse feedback that limits trajectory-level control. We propose a unified framework CogniEdit, combining multi-modal reasoning with dense reward optimization that propagates gradients across consecutive denoising steps, enabling trajectory-level gradient flow through the sampling process. Our method comprises three components: (1) Multi-modal Large Language Models for decomposing complex instructions into actionable directives, (2) Dynamic Token Focus Relocation that adaptively emphasizes fine-grained attributes, and (3) Dense GRPO-based optimization that propagates gradients across consecutive steps for trajectory-level supervision. Extensive experiments on benchmark datasets demonstrate that our CogniEdit achieves state-of-the-art performance in balancing fine-grained instruction following with visual quality and editability preservation
Authors: Jiaqi Wang, Weijia Wu, Yi Zhan, Rui Zhao, Ming Hu, James Cheng, Wei Liu, Philip Torr, Kevin Qinghong Lin
Abstract: Recent advances in video generation have produced vivid content that are often indistinguishable from real videos, making AI-generated video detection an emerging societal challenge. Prior AIGC detection benchmarks mostly evaluate video without audio, target broad narrative domains, and focus on classification solely. Yet it remains unclear whether state-of-the-art video generation models can produce immersive, audio-paired videos that reliably deceive humans and VLMs. To this end, we introduce Video Reality Test, an ASMR-sourced video benchmark suite for testing perceptual realism under tight audio-visual coupling, featuring the following dimensions: \textbf{(i) Immersive ASMR video-audio sources.} Built on carefully curated real ASMR videos, the benchmark targets fine-grained action-object interactions with diversity across objects, actions, and backgrounds. \textbf{(ii) Peer-Review evaluation.} An adversarial creator-reviewer protocol where video generation models act as creators aiming to fool reviewers, while VLMs serve as reviewers seeking to identify fakeness. Our experimental findings show: The best creator Veo3.1-Fast even fools most VLMs: the strongest reviewer (Gemini 2.5-Pro) achieves only 56\% accuracy (random 50\%), far below that of human experts (81.25\%). Adding audio improves real-fake discrimination, yet superficial cues such as watermarks can still significantly mislead models. These findings delineate the current boundary of video generation realism and expose limitations of VLMs in perceptual fidelity and audio-visual consistency. Our code is available at https://github.com/video-reality-test/video-reality-test.
URLs: https://github.com/video-reality-test/video-reality-test.
Authors: Bo Liu, Qiao Qin, Qinghui He
Abstract: The rapid advancement of generative models has increased the demand for generated image detectors capable of generalizing across diverse and evolving generation techniques. However, existing methods, including those leveraging pre-trained vision-language models, often produce highly entangled representations, mixing task-relevant forensic cues (causal features) with spurious or irrelevant patterns (non-causal features), thus limiting generalization. To address this issue, we propose CausalCLIP, a framework that explicitly disentangles causal from non-causal features and employs targeted filtering guided by causal inference principles to retain only the most transferable and discriminative forensic cues. By modeling the generation process with a structural causal model and enforcing statistical independence through Gumbel-Softmax-based feature masking and Hilbert-Schmidt Independence Criterion (HSIC) constraints, CausalCLIP isolates stable causal features robust to distribution shifts. When tested on unseen generative models from different series, CausalCLIP demonstrates strong generalization ability, achieving improvements of 6.83% in accuracy and 4.06% in average precision over state-of-the-art methods.
Authors: Shu Yu, Chaochao Lu
Abstract: Diffusion models (DMs) have achieved remarkable success in image and video generation. However, they still struggle with (1) physical alignment and (2) out-of-distribution (OOD) instruction following. We argue that these issues stem from the models' failure to learn causal directions and to disentangle causal factors for novel recombination. We introduce the Causal Scene Graph (CSG) and the Physical Alignment Probe (PAP) dataset to enable diagnostic interventions. This analysis yields three key insights. First, DMs struggle with multi-hop reasoning for elements not explicitly determined in the prompt. Second, the prompt embedding contains disentangled representations for texture and physics. Third, visual causal structure is disproportionately established during the initial, computationally limited denoising steps. Based on these findings, we introduce LINA (Learning INterventions Adaptively), a novel framework that learns to predict prompt-specific interventions, which employs (1) targeted guidance in the prompt and visual latent spaces, and (2) a reallocated, causality-aware denoising schedule. Our approach enforces both physical alignment and OOD instruction following in image and video DMs, achieving state-of-the-art performance on challenging causal generation tasks and the Winoground dataset. Our project page is at https://opencausalab.github.io/LINA.
Authors: Zhihang Liu, Xiaoyi Bao, Pandeng Li, Junjie Zhou, Zhaohe Liao, Yefei He, Kaixun Jiang, Chen-Wei Xie, Yun Zheng, Hongtao Xie
Abstract: While existing generation and unified models excel at general image generation, they struggle with tasks requiring deep reasoning, planning, and precise data-to-visual mapping abilities beyond general scenarios. To push beyond the existing limitations, we introduce a new and challenging task: creative table visualization, requiring the model to generate an infographic that faithfully and aesthetically visualizes the data from a given table. To address this challenge, we propose ShowTable, a pipeline that synergizes MLLMs with diffusion models via a progressive self-correcting process. The MLLM acts as the central orchestrator for reasoning the visual plan and judging visual errors to provide refined instructions, the diffusion execute the commands from MLLM, achieving high-fidelity results. To support this task and our pipeline, we introduce three automated data construction pipelines for training different modules. Furthermore, we introduce TableVisBench, a new benchmark with 800 challenging instances across 5 evaluation dimensions, to assess performance on this task. Experiments demonstrate that our pipeline, instantiated with different models, significantly outperforms baselines, highlighting its effective multi-modal reasoning, generation, and error correction capabilities.
Authors: Kling Team, Jialu Chen, Yikang Ding, Zhixue Fang, Kun Gai, Yuan Gao, Kang He, Jingyun Hua, Boyuan Jiang, Mingming Lao, Xiaohan Li, Hui Liu, Jiwen Liu, Xiaoqiang Liu, Yuan Liu, Shun Lu, Yongsen Mao, Yingchao Shao, Huafeng Shi, Xiaoyu Shi, Peiqin Sun, Songlin Tang, Pengfei Wan, Chao Wang, Xuebo Wang, Haoxian Zhang, Yuanxing Zhang, Yan Zhou
Abstract: Avatar video generation models have achieved remarkable progress in recent years. However, prior work exhibits limited efficiency in generating long-duration high-resolution videos, suffering from temporal drifting, quality degradation, and weak prompt following as video length increases. To address these challenges, we propose KlingAvatar 2.0, a spatio-temporal cascade framework that performs upscaling in both spatial resolution and temporal dimension. The framework first generates low-resolution blueprint video keyframes that capture global semantics and motion, and then refines them into high-resolution, temporally coherent sub-clips using a first-last frame strategy, while retaining smooth temporal transitions in long-form videos. To enhance cross-modal instruction fusion and alignment in extended videos, we introduce a Co-Reasoning Director composed of three modality-specific large language model (LLM) experts. These experts reason about modality priorities and infer underlying user intent, converting inputs into detailed storylines through multi-turn dialogue. A Negative Director further refines negative prompts to improve instruction alignment. Building on these components, we extend the framework to support ID-specific multi-character control. Extensive experiments demonstrate that our model effectively addresses the challenges of efficient, multimodally aligned long-form high-resolution video generation, delivering enhanced visual clarity, realistic lip-teeth rendering with accurate lip synchronization, strong identity preservation, and coherent multimodal instruction following.
Authors: Mikhail Zakharov
Abstract: Face recognition systems rely on learning highly discriminative and compact identity clusters to enable accurate retrieval. However, as with other surveillance-oriented technologies, such systems raise serious privacy concerns due to their potential for unauthorized identity tracking. While several works have explored machine unlearning as a means of privacy protection, their applicability to face retrieval - especially for modern embedding-based recognition models - remains largely unexplored. In this work, we study the problem of face identity unlearning for retrieval systems and present its inherent challenges. The goal is to make selected identities unretrievable by dispersing their embeddings on the hypersphere and preventing the formation of compact identity clusters that enable re-identification in the gallery. The primary challenge is to achieve this forgetting effect while preserving the discriminative structure of the embedding space and the retrieval performance of the model for the remaining identities. To address this, we evaluate several existing approximate class unlearning methods (e.g., Random Labeling, Gradient Ascent, Boundary Unlearning, and other recent approaches) in the context of face retrieval and propose a simple yet effective dispersion-based unlearning approach. Extensive experiments on standard benchmarks (VGGFace2, CelebA) demonstrate that our method achieves superior forgetting behavior while preserving retrieval utility.
Authors: Elizaveta Prozorova, Anton Konev, Vladimir Faerman
Abstract: The article analyzes the use of thermal imaging technologies for biometric identification based on facial thermograms. It presents a comparative analysis of infrared spectral ranges (NIR, SWIR, MWIR, and LWIR). The paper also defines key requirements for thermal cameras used in biometric systems, including sensor resolution, thermal sensitivity, and a frame rate of at least 30 Hz. Siamese neural networks are proposed as an effective approach for automating the identification process. In experiments conducted on a proprietary dataset, the proposed method achieved an accuracy of approximately 80%. The study also examines the potential of hybrid systems that combine visible and infrared spectra to overcome the limitations of individual modalities. The results indicate that thermal imaging is a promising technology for developing reliable security systems.
Authors: Carla Monteiro, Valentina Corbetta, Regina Beets-Tan, Lu\'is F. Teixeira, Wilson Silva
Abstract: Automatic polyp segmentation is crucial for improving the clinical identification of colorectal cancer (CRC). While Deep Learning (DL) techniques have been extensively researched for this problem, current methods frequently struggle with generalization, particularly in data-constrained or challenging settings. Moreover, many existing polyp segmentation methods rely on complex, task-specific architectures. To address these limitations, we present a framework that leverages the intrinsic robustness of DINO self-attention "key" features for robust segmentation. Unlike traditional methods that extract tokens from the deepest layers of the Vision Transformer (ViT), our approach leverages the key features of the self-attention module with a simple convolutional decoder to predict polyp masks, resulting in enhanced performance and better generalizability. We validate our approach using a multi-center dataset under two rigorous protocols: Domain Generalization (DG) and Extreme Single Domain Generalization (ESDG). Our results, supported by a comprehensive statistical analysis, demonstrate that this pipeline achieves state-of-the-art (SOTA) performance, significantly enhancing generalization, particularly in data-scarce and challenging scenarios. While avoiding a polyp-specific architecture, we surpass well-established models like nnU-Net and UM-Net. Additionally, we provide a systematic benchmark of the DINO framework's evolution, quantifying the specific impact of architectural advancements on downstream polyp segmentation performance.
Authors: Anran Qi, Changjian Li, Adrien Bousseau, Niloy J. Mitra
Abstract: We address image-to-video generation with explicit user control over the final frame's disoccluded regions. Current image-to-video pipelines produce plausible motion but struggle to generate predictable, articulated motions while enforcing user-specified content in newly revealed areas. Our key idea is to separate motion specification from appearance synthesis: we introduce a lightweight, user-editable Proxy Dynamic Graph (PDG) that deterministically yet approximately drives part motion, while a frozen diffusion prior is used to synthesize plausible appearance that follows that motion. In our training-free pipeline, the user loosely annotates and reposes a PDG, from which we compute a dense motion flow to leverage diffusion as a motion-guided shader. We then let the user edit appearance in the disoccluded areas of the image, and exploit the visibility information encoded by the PDG to perform a latent-space composite that reconciles motion with user intent in these areas. This design yields controllable articulation and user control over disocclusions without fine-tuning. We demonstrate clear advantages against state-of-the-art alternatives towards images turned into short videos of articulated objects, furniture, vehicles, and deformables. Our method mixes generative control, in the form of loose pose and structure, with predictable controls, in the form of appearance specification in the final frame in the disoccluded regions, unlocking a new image-to-video workflow. Code will be released on acceptance. Project page: https://anranqi.github.io/beyondvisible.github.io/
Authors: Malte Silbernagel, Albert Alonso, Jens Petersen, Bulat Ibragimov, Marleen de Bruijne, Madeleine K. Wyburd
Abstract: Accurately predicting topologically correct masks remains a difficult task for general segmentation models, which often produce fragmented or disconnected outputs. Fixing these artifacts typically requires hand-crafted refinement rules or architectures specialized to a particular task. Here, we show that Neural Cellular Automata (NCA) can be directly re-purposed as an effective refinement mechanism, using local, iterative updates guided by image context to repair segmentation masks. By training on imperfect masks and ground truths, the automaton learns the structural properties of the target shape while relying solely on local information. When applied to coarse, globally predicted masks, the learned dynamics progressively reconnect broken regions, prune loose fragments and converge towards stable, topologically consistent results. We show how refinement NCA (rNCA) can be easily applied to repair common topological errors produced by different base segmentation models and tasks: for fragmented retinal vessels, it yields 2-3% gains in Dice/clDice and improves Betti errors, reducing $\beta_0$ errors by 60% and $\beta_1$ by 20%; for myocardium, it repairs 61.5% of broken cases in a zero-shot setting while lowering ASSD and HD by 19% and 16%, respectively. This showcases NCA as effective and broadly applicable refiners.
Authors: Lorenzo Pettinari, Sidaty El Hadramy, Michael Wehrli, Philippe C. Cattin, Daniel Studer, Carol C. Hasler, Maria Licci
Abstract: Purpose: Intraoperative navigation in spine surgery demands millimeter-level accuracy. Current systems based on intraoperative radiographic imaging and bone-anchored markers are invasive, radiation-intensive and workflow disruptive. Recent markerless RGB-D registration methods offer a promising alternative, but existing approaches rely on weak segmentation labels to isolate relevant anatomical structures, which can propagate errors throughout registration. Methods: We present End2Reg an end-to-end deep learning framework that jointly optimizes segmentation and registration, eliminating the need for weak segmentation labels and manual steps. The network learns segmentation masks specifically optimized for registration, guided solely by the registration objective without direct segmentation supervision. Results: The proposed framework achieves state-of-the-art performance on ex- and in-vivo benchmarks, reducing median Target Registration Error by 32% to 1.83mm and mean Root Mean Square Error by 45% to 3.95mm, respectively. An ablation study confirms that end-to-end optimization significantly improves registration accuracy. Conclusion: The presented end-to-end RGB-D registration pipeline removes dependency on weak labels and manual steps, advancing towards fully automatic, markerless intraoperative navigation. Code and interactive visualizations are available at: https://lorenzopettinari.github.io/end-2-reg/.
Authors: Patryk Ni\.zeniec, Marcin Iwanowski
Abstract: This paper introduces a novel pipeline for generating large-scale, highly realistic, and automatically labeled datasets for computer vision tasks in robotic environments. Our approach addresses the critical challenges of the domain gap between synthetic and real-world imagery and the time-consuming bottleneck of manual annotation. We leverage 3D Gaussian Splatting (3DGS) to create photorealistic representations of the operational environment and objects. These assets are then used in a game engine where physics simulations create natural arrangements. A novel, two-pass rendering technique combines the realism of splats with a shadow map generated from proxy meshes. This map is then algorithmically composited with the image to add both physically plausible shadows and subtle highlights, significantly enhancing realism. Pixel-perfect segmentation masks are generated automatically and formatted for direct use with object detection models like YOLO. Our experiments show that a hybrid training strategy, combining a small set of real images with a large volume of our synthetic data, yields the best detection and segmentation performance, confirming this as an optimal strategy for efficiently achieving robust and accurate models.
Authors: Ahmed Abul Hasanaath, Hamzah Luqman
Abstract: Continuous sign language recognition (CSLR) requires precise spatio-temporal modeling to accurately recognize sequences of gestures in videos. Existing frameworks often rely on CNN-based spatial backbones combined with temporal convolution or recurrent modules. These techniques fail in capturing fine-grained hand and facial cues and modeling long-range temporal dependencies. To address these limitations, we propose the Unified Spatio-Temporal Modeling (USTM) framework, a spatio-temporal encoder that effectively models complex patterns using a combination of a Swin Transformer backbone enhanced with lightweight temporal adapter with positional embeddings (TAPE). Our framework captures fine-grained spatial features alongside short and long-term temporal context, enabling robust sign language recognition from RGB videos without relying on multi-stream inputs or auxiliary modalities. Extensive experiments on benchmarked datasets including PHOENIX14, PHOENIX14T, and CSL-Daily demonstrate that USTM achieves state-of-the-art performance against RGB-based as well as multi-modal CSLR approaches, while maintaining competitive performance against multi-stream approaches. These results highlight the strength and efficacy of the USTM framework for CSLR. The code is available at https://github.com/gufranSabri/USTM
Authors: Haoxuan Qu, Qiuchi Xiang, Yujun Cai, Yirui Wu, Majid Mirmehdi, Hossein Rahmani, Jun Liu
Abstract: Unexploitable example generation aims to transform personal images into their unexploitable (unlearnable) versions before they are uploaded online, thereby preventing unauthorized exploitation of online personal images. Recently, this task has garnered significant research attention due to its critical relevance to personal data privacy. Yet, despite recent progress, existing methods for this task can still suffer from limited practical applicability, as they can fail to generate examples that are broadly unexploitable across different real-world computer vision tasks. To deal with this problem, in this work, we propose a novel Meta Cross-Task Unexploitable Example Generation (MCT-UEG) framework. At the core of our framework, to optimize the unexploitable example generator for effectively producing broadly unexploitable examples, we design a flat-minima-oriented meta training and testing scheme. Extensive experiments show the efficacy of our framework.
Authors: Qingyu Shi, Size Wu, Jinbin Bai, Kaidong Yu, Yujing Wang, Yunhai Tong, Xiangtai Li, Xuelong Li
Abstract: Visual tokenizers play a crucial role in diffusion models. The dimensionality of latent space governs both reconstruction fidelity and the semantic expressiveness of the latent feature. However, a fundamental trade-off is inherent between dimensionality and generation quality, constraining existing methods to low-dimensional latent spaces. Although recent works have leveraged vision foundation models to enrich the semantics of visual tokenizers and accelerate convergence, high-dimensional tokenizers still underperform their low-dimensional counterparts. In this work, we propose RecTok, which overcomes the limitations of high-dimensional visual tokenizers through two key innovations: flow semantic distillation and reconstruction--alignment distillation. Our key insight is to make the forward flow in flow matching semantically rich, which serves as the training space of diffusion transformers, rather than focusing on the latent space as in previous works. Specifically, our method distills the semantic information in VFMs into the forward flow trajectories in flow matching. And we further enhance the semantics by introducing a masked feature reconstruction loss. Our RecTok achieves superior image reconstruction, generation quality, and discriminative performance. It achieves state-of-the-art results on the gFID-50K under both with and without classifier-free guidance settings, while maintaining a semantically rich latent space structure. Furthermore, as the latent dimensionality increases, we observe consistent improvements. Code and model are available at https://shi-qingyu.github.io/rectok.github.io.
Authors: Noa Cohen, Nurit Spingarn-Eliezer, Inbar Huberman-Spiegelglas, Tomer Michaeli
Abstract: Text-to-Image (TTI) models generate images based on text prompts, which often leave certain aspects of the desired image ambiguous. When faced with these ambiguities, TTI models have been shown to exhibit biases in their interpretations. These biases can have societal impacts, e.g., when showing only a certain race for a stated occupation. They can also affect user experience when creating redundancy within a set of generated images instead of spanning diverse possibilities. Here, we introduce MineTheGap - a method for automatically mining prompts that cause a TTI model to generate biased outputs. Our method goes beyond merely detecting bias for a given prompt. Rather, it leverages a genetic algorithm to iteratively refine a pool of prompts, seeking for those that expose biases. This optimization process is driven by a novel bias score, which ranks biases according to their severity, as we validate on a dataset with known biases. For a given prompt, this score is obtained by comparing the distribution of generated images to the distribution of LLM-generated texts that constitute variations on the prompt. Code and examples are available on the project's webpage.
Authors: Anika Islam, Tasfia Tahsin, Zaarin Anjum, Md. Bakhtiar Hasan, Md. Hasanul Kabir
Abstract: Accurate and timely identification of plant leaf diseases is essential for resilient and sustainable agriculture, yet most deep learning approaches rely on large annotated datasets and computationally intensive models that are unsuitable for data-scarce and resource-constrained environments. To address these challenges we present a few-shot learning approach within a lightweight yet efficient framework that combines domain-adapted MobileNetV2 and MobileNetV3 models as feature extractors, along with a feature fusion technique to generate robust feature representation. For the classification task, the fused features are passed through a Bi-LSTM classifier enhanced with attention mechanisms to capture sequential dependencies and focus on the most relevant features, thereby achieving optimal classification performance even in complex, real-world environments with noisy or cluttered backgrounds. The proposed framework was evaluated across multiple experimental setups, including both laboratory-controlled and field-captured datasets. On tomato leaf diseases from the PlantVillage dataset, it consistently improved performance across 1 to 15 shot scenarios, reaching 98.23+-0.33% at 15 shot, closely approaching the 99.98% SOTA benchmark achieved by a Transductive LSTM with attention, while remaining lightweight and mobile-friendly. Under real-world conditions using field images from the Dhan Shomadhan dataset, it maintained robust performance, reaching 69.28+-1.49% at 15-shot and demonstrating strong resilience to complex backgrounds. Notably, it also outperformed the previous SOTA accuracy of 96.0% on six diseases from PlantVillage, achieving 99.72% with only 15-shot learning. With a compact model size of approximately 40 MB and inference complexity of approximately 1.12 GFLOPs, this work establishes a scalable, mobile-ready foundation for precise plant disease diagnostics in data-scarce regions.
Authors: Thalyssa Baiocco-Rodrigues, Antoine Olivier, Reda Belbahri, Thomas Duboudin, Pierre-Antoine Bannier, Benjamin Adjadj, Katharina Von Loga, Nathan Noiry, Maxime Touzot, Hector Roux de Bezieux
Abstract: As the therapeutic target for Inflammatory Bowel Disease (IBD) shifts toward histologic remission, the accurate assessment of microscopic inflammation has become increasingly central for evaluating disease activity and response to treatment. In this work, we introduce IMILIA (Interpretable Multiple Instance Learning for Inflammation Analysis), an end-to-end framework designed for the prediction of inflammation presence in IBD digitized slides stained with hematoxylin and eosin (H&E), followed by the automated computation of markers characterizing tissue regions driving the predictions. IMILIA is composed of an inflammation prediction module, consisting of a Multiple Instance Learning (MIL) model, and an interpretability module, divided in two blocks: HistoPLUS, for cell instance detection, segmentation and classification; and EpiSeg, for epithelium segmentation. IMILIA achieves a cross-validation ROC-AUC of 0.83 on the discovery cohort, and a ROC-AUC of 0.99 and 0.84 on two external validation cohorts. The interpretability module yields biologically consistent insights: tiles with higher predicted scores show increased densities of immune cells (lymphocytes, plasmocytes, neutrophils and eosinophils), whereas lower-scored tiles predominantly contain normal epithelial cells. Notably, these patterns were consistent across all datasets. Code and models to partially replicate the results on the public IBDColEpi dataset can be found at https://github.com/owkin/imilia.
Authors: Arpit Jadon, Joshua Niemeijer, Yuki M. Asano
Abstract: Generative foundation models contain broad visual knowledge and can produce diverse image variations, making them particularly promising for advancing domain generalization tasks. While they can be used for training data augmentation, synthesizing comprehensive target-domain variations remains slow, expensive, and incomplete. We propose an alternative: using diffusion models at test time to map target images back to the source distribution where the downstream model was trained. This approach requires only a source domain description, preserves the task model, and eliminates large-scale synthetic data generation. We demonstrate consistent improvements across segmentation, detection, and classification tasks under challenging environmental shifts in real-to-real domain generalization scenarios with unknown target distributions. Our analysis spans multiple generative and downstream models, including an ensemble variant for enhanced robustness. The method achieves substantial relative gains: 137% on BDD100K-Night, 68% on ImageNet-R, and 62% on DarkZurich.
Authors: Ruiyan Wang, Teng Hu, Kaihui Huang, Zihan Su, Ran Yi, Lizhuang Ma
Abstract: Pose-guided video generation refers to controlling the motion of subjects in generated video through a sequence of poses. It enables precise control over subject motion and has important applications in animation. However, current pose-guided video generation methods are limited to accepting only human poses as input, thus generalizing poorly to pose of other subjects. To address this issue, we propose PoseAnything, the first universal pose-guided video generation framework capable of handling both human and non-human characters, supporting arbitrary skeletal inputs. To enhance consistency preservation during motion, we introduce Part-aware Temporal Coherence Module, which divides the subject into different parts, establishes part correspondences, and computes cross-attention between corresponding parts across frames to achieve fine-grained part-level consistency. Additionally, we propose Subject and Camera Motion Decoupled CFG, a novel guidance strategy that, for the first time, enables independent camera movement control in pose-guided video generation, by separately injecting subject and camera motion control information into the positive and negative anchors of CFG. Furthermore, we present XPose, a high-quality public dataset containing 50,000 non-human pose-video pairs, along with an automated pipeline for annotation and filtering. Extensive experiments demonstrate that Pose-Anything significantly outperforms state-of-the-art methods in both effectiveness and generalization.
Authors: Jiangning Zhang, Junwei Zhu, Teng Hu, Yabiao Wang, Donghao Luo, Weijian Cao, Zhenye Gan, Xiaobin Hu, Zhucun Xue, Chengjie Wang
Abstract: Native 4K (2160$\times$3840) video generation remains a critical challenge due to the quadratic computational explosion of full-attention as spatiotemporal resolution increases, making it difficult for models to strike a balance between efficiency and quality. This paper proposes a novel Transformer retrofit strategy termed $\textbf{T3}$ ($\textbf{T}$ransform $\textbf{T}$rained $\textbf{T}$ransformer) that, without altering the core architecture of full-attention pretrained models, significantly reduces compute requirements by optimizing their forward logic. Specifically, $\textbf{T3-Video}$ introduces a multi-scale weight-sharing window attention mechanism and, via hierarchical blocking together with an axis-preserving full-attention design, can effect an "attention pattern" transformation of a pretrained model using only modest compute and data. Results on 4K-VBench show that $\textbf{T3-Video}$ substantially outperforms existing approaches: while delivering performance improvements (+4.29$\uparrow$ VQA and +0.08$\uparrow$ VTC), it accelerates native 4K video generation by more than 10$\times$. Project page at https://zhangzjn.github.io/projects/T3-Video
Authors: Jiangning Zhang, Junwei Zhu, Zhenye Gan, Donghao Luo, Chuming Lin, Feifan Xu, Xu Peng, Jianlong Hu, Yuansen Liu, Yijia Hong, Weijian Cao, Han Feng, Xu Chen, Chencan Fu, Keke He, Xiaobin Hu, Chengjie Wang
Abstract: We propose a multimodal-driven framework for high-fidelity long-term digital human animation termed $\textbf{Soul}$, which generates semantically coherent videos from a single-frame portrait image, text prompts, and audio, achieving precise lip synchronization, vivid facial expressions, and robust identity preservation. We construct Soul-1M, containing 1 million finely annotated samples with a precise automated annotation pipeline (covering portrait, upper-body, full-body, and multi-person scenes) to mitigate data scarcity, and we carefully curate Soul-Bench for comprehensive and fair evaluation of audio-/text-guided animation methods. The model is built on the Wan2.2-5B backbone, integrating audio-injection layers and multiple training strategies together with threshold-aware codebook replacement to ensure long-term generation consistency. Meanwhile, step/CFG distillation and a lightweight VAE are used to optimize inference efficiency, achieving an 11.4$\times$ speedup with negligible quality loss. Extensive experiments show that Soul significantly outperforms current leading open-source and commercial models on video quality, video-text alignment, identity preservation, and lip-synchronization accuracy, demonstrating broad applicability in real-world scenarios such as virtual anchors and film production. Project page at https://zhangzjn.github.io/projects/Soul/
Authors: Siyan Chen, Yanfei Chen, Ying Chen, Zhuo Chen, Feng Cheng, Xuyan Chi, Jian Cong, Qinpeng Cui, Qide Dong, Junliang Fan, Jing Fang, Zetao Fang, Chengjian Feng, Han Feng, Mingyuan Gao, Yu Gao, Qiushan Guo, Boyang Hao, Qingkai Hao, Bibo He, Qian He, Tuyen Hoang, Ruoqing Hu, Xi Hu, Weilin Huang, Zhaoyang Huang, Zhongyi Huang, Siqi Jiang, Wei Jiang, Yunpu Jiang, Zhuo Jiang, Ashley Kim, Jianan Kong, Zhichao Lai, Shanshan Lao, Ai Li, Feiya Li, Gen Li, Huixia Li, JiaShi Li, Liang Li, Ming Li, Tao Li, Xian Li, Xiaojie Li, Xiaoyang Li, Xingxing Li, Yameng Li, Yifu Li, Yiying Li, Chao Liang, Ying Liang, Zhiqiang Liang, Wang Liao, Yalin Liao, Heng Lin, Kengyu Lin, Shanchuan Lin, Xi Lin, Zhijie Lin, Feng Ling, Fangfang Liu, Gaohong Liu, Jiawei Liu, Jie Liu, Shouda Liu, Shu Liu, Sichao Liu, Songwei Liu, Xin Liu, Xue Liu, Yibo Liu, Zikun Liu, Zuxi Liu, Junlin Lyu, Lecheng Lyu, Qian Lyu, Han Mu, Xiaonan Nie, Jingzhe Ning, Xitong Pan, Yanghua Peng, Lianke Qin, Xueqiong Qu, Yuxi Ren, Yuchen Shen, Guang Shi, Lei Shi, Yan Song, Yinglong Song, Fan Sun, Li Sun, Renfei Sun, Zeyu Sun, Wenjing Tang, Zirui Tao, Feng Wang, Furui Wang, Jinran Wang, Junkai Wang, Ke Wang, Kexin Wang, Qingyi Wang, Rui Wang, Sen Wang, Shuai Wang, Tingru Wang, Weichen Wang, Xin Wang, Yanhui Wang, Yue Wang, Yuping Wang, Yuxuan Wang, Ziyu Wang, Guoqiang Wei, Wanru Wei, Di Wu, Guohong Wu, Hanjie Wu, Jian Wu, Jie Wu, Ruolan Wu, Xinglong Wu, Yonghui Wu, Ruiqi Xia, Liang Xiang, Fei Xiao, XueFeng Xiao, Pan Xie, Shuangyi Xie, Shuang Xu, Jinlan Xue, Bangbang Yang, Ceyuan Yang, Jiaqi Yang, Runkai Yang, Tao Yang, Yang Yang, Yihang Yang, ZhiXian Yang, Ziyan Yang, Yifan Yao, Zilyu Ye, Bowen Yu, Chujie Yuan, Linxiao Yuan, Sichun Zeng, Weihong Zeng, Xuejiao Zeng, Yan Zeng, Chuntao Zhang, Heng Zhang, Jingjie Zhang, Kuo Zhang, Liang Zhang, Liying Zhang, Manlin Zhang, Ting Zhang, Weida Zhang, Xiaohe Zhang, Xinyan Zhang, Yan Zhang, Yuan Zhang, Zixiang Zhang, Fengxuan Zhao, Huating Zhao, Yang Zhao, Hao Zheng, Jianbin Zheng, Xiaozheng Zheng, Yangyang Zheng, Yijie Zheng, Jiexin Zhou, Kuan Zhu, Shenhan Zhu, Wenjia Zhu, Benhui Zou, Feilong Zuo
Abstract: Recent strides in video generation have paved the way for unified audio-visual generation. In this work, we present Seedance 1.5 pro, a foundational model engineered specifically for native, joint audio-video generation. Leveraging a dual-branch Diffusion Transformer architecture, the model integrates a cross-modal joint module with a specialized multi-stage data pipeline, achieving exceptional audio-visual synchronization and superior generation quality. To ensure practical utility, we implement meticulous post-training optimizations, including Supervised Fine-Tuning (SFT) on high-quality datasets and Reinforcement Learning from Human Feedback (RLHF) with multi-dimensional reward models. Furthermore, we introduce an acceleration framework that boosts inference speed by over 10X. Seedance 1.5 pro distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation. Seedance 1.5 pro is now accessible on Volcano Engine at https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.
URLs: https://console.volcengine.com/ark/region:ark+cn-beijing/experience/vision?type=GenVideo.
Authors: Piyush Bagad, Andrew Zisserman
Abstract: Our objective is to build a general time-aware video-text embedding model for retrieval. To that end, we propose a simple and efficient recipe, dubbed TARA (Time Aware Retrieval Adaptation), to adapt Multimodal LLMs (MLLMs) to a time-aware video-text embedding model without using any video data at all. For evaluating time-awareness in retrieval, we propose a new benchmark with temporally opposite (chiral) actions as hard negatives and curated splits for chiral and non-chiral actions. We show that TARA outperforms all existing video-text models on this chiral benchmark while also achieving strong results on standard benchmarks. Furthermore, we discover additional benefits of TARA beyond time-awareness: (i) TARA embeddings are negation-aware as shown in NegBench benchmark that evaluates negation in video retrieval, (ii) TARA achieves state of the art performance on verb and adverb understanding in videos. Overall, TARA yields a strong, versatile, time-aware video-text embedding model with state of the art zero-shot performance.
Authors: Marianne Rakic, Siyu Gai, Etienne Chollet, John V. Guttag, Adrian V. Dalca
Abstract: A single biomedical image can be meaningfully segmented in multiple ways, depending on the desired application. For instance, a brain MRI can be segmented according to tissue types, vascular territories, broad anatomical regions, fine-grained anatomy, or pathology, etc. Existing automatic segmentation models typically either (1) support only a single protocol, the one they were trained on, or (2) require labor-intensive manual prompting to specify the desired segmentation. We introduce Pancakes, a framework that, given a new image from a previously unseen domain, automatically generates multi-label segmentation maps for multiple plausible protocols, while maintaining semantic consistency across related images. Pancakes introduces a new problem formulation that is not currently attainable by existing foundation models. In a series of experiments on seven held-out datasets, we demonstrate that our model can significantly outperform existing foundation models in producing several plausible whole-image segmentations, that are semantically coherent across images.
Authors: Shun Maeda, Chunzhi Gu, Koichiro Kamide, Katsuya Hotta, Shangce Gao, Chao Zhang
Abstract: Human-centric anomaly detection (AD) has been primarily studied to specify anomalous behaviors in a single person. However, as humans by nature tend to act in a collaborative manner, behavioral anomalies can also arise from human-human interactions. Detecting such anomalies using existing single-person AD models is prone to low accuracy, as these approaches are typically not designed to capture the complex and asymmetric dynamics of interactions. In this paper, we introduce a novel task, Human-Human Interaction Anomaly Detection (H2IAD), which aims to identify anomalous interactive behaviors within collaborative 3D human actions. To address H2IAD, we then propose Interaction Anomaly Detection Network (IADNet), which is formalized with a Temporal Attention Sharing Module (TASM). Specifically, in designing TASM, we share the encoded motion embeddings across both people such that collaborative motion correlations can be effectively synchronized. Moreover, we notice that in addition to temporal dynamics, human interactions are also characterized by spatial configurations between two people. We thus introduce a Distance-Based Relational Encoding Module (DREM) to better reflect social cues in H2IAD. The normalizing flow is eventually employed for anomaly scoring. Extensive experiments on human-human motion benchmarks demonstrate that IADNet outperforms existing Human-centric AD baselines in H2IAD.
Authors: Tao Zhang, Ziqi Zhang, Zongyang Ma, Yuxin Chen, Bing Li, Chunfeng Yuan, Guangting Wang, Fengyun Rao, Ying Shan, Weiming Hu
Abstract: The ability to perform multi-modal multi-hop reasoning by iteratively integrating information across various modalities and external knowledge is critical for addressing complex real-world challenges. However, existing Multi-modal Large Language Models (MLLMs) are predominantly limited to single-step reasoning, as existing benchmarks lack the complexity needed to evaluate and drive multi-hop abilities. To bridge this gap, we introduce MMhops, a novel, large-scale benchmark designed to systematically evaluate and foster multi-modal multi-hop reasoning. MMhops dataset comprises two challenging task formats, Bridging and Comparison, which necessitate that models dynamically construct complex reasoning chains by integrating external knowledge. To tackle the challenges posed by MMhops, we propose MMhops-R1, a novel multi-modal Retrieval-Augmented Generation (mRAG) framework for dynamic reasoning. Our framework utilizes reinforcement learning to optimize the model for autonomously planning reasoning paths, formulating targeted queries, and synthesizing multi-level information. Comprehensive experiments demonstrate that MMhops-R1 significantly outperforms strong baselines on MMhops, highlighting that dynamic planning and multi-modal knowledge integration are crucial for complex reasoning. Moreover, MMhops-R1 demonstrates strong generalization to tasks requiring fixed-hop reasoning, underscoring the robustness of our dynamic planning approach. In conclusion, our work contributes a challenging new benchmark and a powerful baseline model, and we will release the associated code, data, and weights to catalyze future research in this critical area.
Authors: Christophe Bolduc, Julien Philip, Li Ma, Mingming He, Paul Debevec, Jean-Fran\c{c}ois Lalonde
Abstract: We present Lighting in Motion (LiMo), a diffusion-based approach to spatiotemporal lighting estimation. LiMo targets both realistic high-frequency detail prediction and accurate illuminance estimation. To account for both, we propose generating a set of mirrored and diffuse spheres at different exposures, based on their 3D positions in the input. Making use of diffusion priors, we fine-tune powerful existing diffusion models on a large-scale customized dataset of indoor and outdoor scenes, paired with spatiotemporal light probes. For accurate spatial conditioning, we demonstrate that depth alone is insufficient and we introduce a new geometric condition to provide the relative position of the scene to the target 3D position. Finally, we combine diffuse and mirror predictions at different exposures into a single HDRI map leveraging differentiable rendering. We thoroughly evaluate our method and design choices to establish LiMo as state-of-the-art for both spatial control and prediction accuracy.
Authors: Haoyue Zhang, Meera Chappidi, Erolcan Sayar, Helen Richards, Zhijun Chen, Lucas Liu, Roxanne Wadia, Peter A Humphrey, Fady Ghali, Alberto Contreras-Sanz, Peter Black, Jonathan Wright, Stephanie Harmon, Michael Haffner
Abstract: Recent deep learning frameworks in histopathology, particularly multiple instance learning (MIL) combined with pathology foundational models (PFMs), have shown strong performance. However, PFMs exhibit limitations on certain cancer or specimen types due to domain shifts - these cancer types were rarely used for pretraining or specimens contain tissue-based artifacts rarely seen within the pretraining population. Such is the case for transurethral resection of bladder tumor (TURBT), which are essential for diagnosing muscle-invasive bladder cancer (MIBC), but contain fragmented tissue chips and electrocautery artifacts and were not widely used in publicly available PFMs. To address this, we propose a simple yet effective domain-adaptive self-supervised adaptor (DA-SSL) that realigns pretrained PFM features to the TURBT domain without fine-tuning the foundational model itself. We pilot this framework for predicting treatment response in TURBT, where histomorphological features are currently underutilized and identifying patients who will benefit from neoadjuvant chemotherapy (NAC) is challenging. In our multi-center study, DA-SSL achieved an AUC of 0.77+/-0.04 in five-fold cross-validation and an external test accuracy of 0.84, sensitivity of 0.71, and specificity of 0.91 using majority voting. Our results demonstrate that lightweight domain adaptation with self-supervision can effectively enhance PFM-based MIL pipelines for clinically challenging histopathology tasks. Code is Available at https://github.com/zhanghaoyue/DA_SSL_TURBT.
Authors: Jianxiong Gao, Zhaoxi Chen, Xian Liu, Junhao Zhuang, Chengming Xu, Jianfeng Feng, Yu Qiao, Yanwei Fu, Chenyang Si, Ziwei Liu
Abstract: Building video world models upon pretrained video generation systems represents an important yet challenging step toward general spatiotemporal intelligence. A world model should possess three essential properties: controllability, long-term visual quality, and temporal consistency. To this end, we take a progressive approach-first enhancing controllability and then extending toward long-term, high-quality generation. We present LongVie 2, an end-to-end autoregressive framework trained in three stages: (1) Multi-modal guidance, which integrates dense and sparse control signals to provide implicit world-level supervision and improve controllability; (2) Degradation-aware training on the input frame, bridging the gap between training and long-term inference to maintain high visual quality; and (3) History-context guidance, which aligns contextual information across adjacent clips to ensure temporal consistency. We further introduce LongVGenBench, a comprehensive benchmark comprising 100 high-resolution one-minute videos covering diverse real-world and synthetic environments. Extensive experiments demonstrate that LongVie 2 achieves state-of-the-art performance in long-range controllability, temporal coherence, and visual fidelity, and supports continuous video generation lasting up to five minutes, marking a significant step toward unified video world modeling.
Authors: Felix J. Dorfner, Manon A. Dorster, Ryan Connolly, Oscar Gentilhomme, Edward Gibbs, Steven Graham, Seth Wander, Thomas Schultz, Manisha Bahl, Dania Daye, Albert E. Kim, Christopher P. Bridge
Abstract: Foundation models have shown promise in medical imaging but remain underexplored for three-dimensional imaging modalities. No foundation model currently exists for Digital Breast Tomosynthesis (DBT), despite its use for breast cancer screening. To develop and evaluate a foundation model for DBT (DBT-DINO) across multiple clinical tasks and assess the impact of domain-specific pre-training. Self-supervised pre-training was performed using the DINOv2 methodology on over 25 million 2D slices from 487,975 DBT volumes from 27,990 patients. Three downstream tasks were evaluated: (1) breast density classification using 5,000 screening exams; (2) 5-year risk of developing breast cancer using 106,417 screening exams; and (3) lesion detection using 393 annotated volumes. For breast density classification, DBT-DINO achieved an accuracy of 0.79 (95\% CI: 0.76--0.81), outperforming both the MetaAI DINOv2 baseline (0.73, 95\% CI: 0.70--0.76, p<.001) and DenseNet-121 (0.74, 95\% CI: 0.71--0.76, p<.001). For 5-year breast cancer risk prediction, DBT-DINO achieved an AUROC of 0.78 (95\% CI: 0.76--0.80) compared to DINOv2's 0.76 (95\% CI: 0.74--0.78, p=.57). For lesion detection, DINOv2 achieved a higher average sensitivity of 0.67 (95\% CI: 0.60--0.74) compared to DBT-DINO with 0.62 (95\% CI: 0.53--0.71, p=.60). DBT-DINO demonstrated better performance on cancerous lesions specifically with a detection rate of 78.8\% compared to Dinov2's 77.3\%. Using a dataset of unprecedented size, we developed DBT-DINO, the first foundation model for DBT. DBT-DINO demonstrated strong performance on breast density classification and cancer risk prediction. However, domain-specific pre-training showed variable benefits on the detection task, with ImageNet baseline outperforming DBT-DINO on general lesion detection, indicating that localized detection tasks require further methodological development.
Authors: Shweta Mahajan, Shreya Kadambi, Hoang Le, Munawar Hayat, Fatih Porikli
Abstract: We introduce the Do-Undo task and benchmark to address a critical gap in vision-language models: understanding and generating physically plausible scene transformations driven by real-world actions. Unlike prior work focused on object-level edits, Do-Undo requires models to simulate the outcome of a physical action and then accurately reverse it, reflecting true cause-and-effect in the visual world. We curate a large-scale dataset of reversible actions from real-world videos and design a training strategy enforcing consistency for robust action grounding. Our experiments reveal that current models struggle with physical reversibility, underscoring the importance of this task for embodied AI, robotics, and physics-aware generative modeling. Do-Undo establishes an intuitive testbed for evaluating and advancing physical reasoning in multimodal systems.
Authors: Junchao Zhu, Ruining Deng, Junlin Guo, Tianyuan Yao, Chongyu Qu, Juming Xiong, Siqi Lu, Zhengyi Lu, Yanfan Zhu, Marilyn Lionts, Yuechen Yang, Yalin Zheng, Yu Wang, Shilin Zhao, Haichun Yang, Yuankai Huo
Abstract: Spatial transcriptomics (ST) is an emerging technology that enables researchers to investigate the molecular relationships underlying tissue morphology. However, acquiring ST data remains prohibitively expensive, and traditional fixed-grid sampling strategies lead to redundant measurements of morphologically similar or biologically uninformative regions, thus resulting in scarce data that constrain current methods. The well-established single-cell sequencing field, however, could provide rich biological data as an effective auxiliary source to mitigate this limitation. To bridge these gaps, we introduce SCR2-ST, a unified framework that leverages single-cell prior knowledge to guide efficient data acquisition and accurate expression prediction. SCR2-ST integrates a single-cell guided reinforcement learning-based (SCRL) active sampling and a hybrid regression-retrieval prediction network SCR2Net. SCRL combines single-cell foundation model embeddings with spatial density information to construct biologically grounded reward signals, enabling selective acquisition of informative tissue regions under constrained sequencing budgets. SCR2Net then leverages the actively sampled data through a hybrid architecture combining regression-based modeling with retrieval-augmented inference, where a majority cell-type filtering mechanism suppresses noisy matches and retrieved expression profiles serve as soft labels for auxiliary supervision. We evaluated SCR2-ST on three public ST datasets, demonstrating SOTA performance in both sampling efficiency and prediction accuracy, particularly under low-budget scenarios. Code is publicly available at: https://github.com/hrlblab/SCR2ST
Authors: Haoyu Fu, Diankun Zhang, Zongchuang Zhao, Jianfeng Cui, Hongwei Xie, Bing Wang, Guang Chen, Dingkang Liang, Xiang Bai
Abstract: Current Vision-Language-Action (VLA) paradigms in autonomous driving primarily rely on Imitation Learning (IL), which introduces inherent challenges such as distribution shift and causal confusion. Online Reinforcement Learning offers a promising pathway to address these issues through trial-and-error learning. However, applying online reinforcement learning to VLA models in autonomous driving is hindered by inefficient exploration in continuous action spaces. To overcome this limitation, we propose MindDrive, a VLA framework comprising a large language model (LLM) with two distinct sets of LoRA parameters. The one LLM serves as a Decision Expert for scenario reasoning and driving decision-making, while the other acts as an Action Expert that dynamically maps linguistic decisions into feasible trajectories. By feeding trajectory-level rewards back into the reasoning space, MindDrive enables trial-and-error learning over a finite set of discrete linguistic driving decisions, instead of operating directly in a continuous action space. This approach effectively balances optimal decision-making in complex scenarios, human-like driving behavior, and efficient exploration in online reinforcement learning. MindDrive achieves strong closed-loop performance on the challenging Bench2Drive benchmark, with a Driving Score (DS) of 78.04 and a Success Rate (SR) of 55.09%. To the best of our knowledge, this is the first work to demonstrate the effectiveness of online reinforcement learning for the VLA model in autonomous driving.
Authors: Michal Nazarczuk, Thomas Tanay, Arthur Moreau, Zhensong Zhang, Eduardo P\'erez-Pellitero
Abstract: This paper presents a new dataset for Novel View Synthesis, generated from a high-quality, animated film with stunning realism and intricate detail. Our dataset captures a variety of dynamic scenes, complete with detailed textures, lighting, and motion, making it ideal for training and evaluating cutting-edge 4D scene reconstruction and novel view generation models. In addition to high-fidelity RGB images, we provide multiple complementary modalities, including depth, surface normals, object segmentation and optical flow, enabling a deeper understanding of scene geometry and motion. The dataset is organised into three distinct benchmarking scenarios: a dense multi-view camera setup, a sparse camera arrangement, and monocular video sequences, enabling a wide range of experimentation and comparison across varying levels of data sparsity. With its combination of visual richness, high-quality annotations, and diverse experimental setups, this dataset offers a unique resource for pushing the boundaries of view synthesis and 3D vision.
Authors: Wenhan Chen, Sezer Karaoglu, Theo Gevers
Abstract: Recent advances in diffusion-based generation techniques enable AI models to produce highly realistic videos, heightening the need for reliable detection mechanisms. However, existing detection methods provide only limited exploration of the 3D geometric patterns present in generated videos. In this paper, we use vanishing points as an explicit representation of 3D geometry patterns, revealing fundamental discrepancies in geometric consistency between real and AI-generated videos. We introduce Grab-3D, a geometry-aware transformer framework for detecting AI-generated videos based on 3D geometric temporal consistency. To enable reliable evaluation, we construct an AI-generated video dataset of static scenes, allowing stable 3D geometric feature extraction. We propose a geometry-aware transformer equipped with geometric positional encoding, temporal-geometric attention, and an EMA-based geometric classifier head to explicitly inject 3D geometric awareness into temporal modeling. Experiments demonstrate that Grab-3D significantly outperforms state-of-the-art detectors, achieving robust cross-domain generalization to unseen generators.
Authors: Junwen Miao, Penghui Du, Yi Liu, Yu Wang, Yan Wang
Abstract: Industrial anomaly detection (IAD) is difficult due to the scarcity of normal reference samples and the subtle, localized nature of many defects. Single-pass vision-language models (VLMs) often overlook small abnormalities and lack explicit mechanisms to compare against canonical normal patterns. We propose AgentIAD, a tool-driven agentic framework that enables multi-stage visual inspection. The agent is equipped with a Perceptive Zoomer (PZ) for localized fine-grained analysis and a Comparative Retriever (CR) for querying normal exemplars when evidence is ambiguous. To teach these inspection behaviors, we construct structured perceptive and comparative trajectories from the MMAD dataset and train the model in two stages: supervised fine-tuning followed by reinforcement learning. A two-part reward design drives this process: a perception reward that supervises classification accuracy, spatial alignment, and type correctness, and a behavior reward that encourages efficient tool use. Together, these components enable the model to refine its judgment through step-wise observation, zooming, and verification. AgentIAD achieves a new state-of-the-art 97.62% classification accuracy on MMAD, surpassing prior MLLM-based approaches while producing transparent and interpretable inspection traces.
Authors: Yiyi Cai, Xuangeng Chu, Xiwei Gao, Sitong Gong, Yifei Huang, Caixin Kang, Kunhang Li, Haiyang Liu, Ruicong Liu, Yun Liu, Dianwen Ng, Zixiong Su, Erwin Wu, Yuhan Wu, Dingkun Yan, Tianyu Yan, Chang Zeng, Bo Zheng, You Zhou
Abstract: We introduce Interactive Intelligence, a novel paradigm of digital human that is capable of personality-aligned expression, adaptive interaction, and self-evolution. To realize this, we present Mio (Multimodal Interactive Omni-Avatar), an end-to-end framework composed of five specialized modules: Thinker, Talker, Face Animator, Body Animator, and Renderer. This unified architecture integrates cognitive reasoning with real-time multimodal embodiment to enable fluid, consistent interaction. Furthermore, we establish a new benchmark to rigorously evaluate the capabilities of interactive intelligence. Extensive experiments demonstrate that our framework achieves superior performance compared to state-of-the-art methods across all evaluated dimensions. Together, these contributions move digital humans beyond superficial imitation toward intelligent interaction.
Authors: Xiaohu Huang, Hao Zhou, Qiangpeng Yang, Shilei Wen, Kai Han
Abstract: In this paper, we present JoVA, a unified framework for joint video-audio generation. Despite recent encouraging advances, existing methods face two critical limitations. First, most existing approaches can only generate ambient sounds and lack the capability to produce human speech synchronized with lip movements. Second, recent attempts at unified human video-audio generation typically rely on explicit fusion or modality-specific alignment modules, which introduce additional architecture design and weaken the model simplicity of the original transformers. To address these issues, JoVA employs joint self-attention across video and audio tokens within each transformer layer, enabling direct and efficient cross-modal interaction without the need for additional alignment modules. Furthermore, to enable high-quality lip-speech synchronization, we introduce a simple yet effective mouth-area loss based on facial keypoint detection, which enhances supervision on the critical mouth region during training without compromising architectural simplicity. Extensive experiments on benchmarks demonstrate that JoVA outperforms or is competitive with both unified and audio-driven state-of-the-art methods in lip-sync accuracy, speech quality, and overall video-audio generation fidelity. Our results establish JoVA as an elegant framework for high-quality multimodal generation.
Authors: Ziqi Ma, Hongqiao Chen, Yisong Yue, Georgia Gkioxari
Abstract: Recent progress in image-to-3D has opened up immense possibilities for design, AR/VR, and robotics. However, to use AI-generated 3D assets in real applications, a critical requirement is the capability to edit them easily. We present a feedforward method, Steer3D, to add text steerability to image-to-3D models, which enables editing of generated 3D assets with language. Our approach is inspired by ControlNet, which we adapt to image-to-3D generation to enable text steering directly in a forward pass. We build a scalable data engine for automatic data generation, and develop a two-stage training recipe based on flow-matching training and Direct Preference Optimization (DPO). Compared to competing methods, Steer3D more faithfully follows the language instruction and maintains better consistency with the original 3D asset, while being 2.4x to 28.5x faster. Steer3D demonstrates that it is possible to add a new modality (text) to steer the generation of pretrained image-to-3D generative models with 100k data. Project website: https://glab-caltech.github.io/steer3d/
Authors: Tianye Ding, Yiming Xie, Yiqing Liang, Moitreya Chatterjee, Pedro Miraldo, Huaizu Jiang
Abstract: Recent feed-forward reconstruction models like VGGT and $\pi^3$ achieve impressive reconstruction quality but cannot process streaming videos due to quadratic memory complexity, limiting their practical deployment. While existing streaming methods address this through learned memory mechanisms or causal attention, they require extensive retraining and may not fully leverage the strong geometric priors of state-of-the-art offline models. We propose LASER, a training-free framework that converts an offline reconstruction model into a streaming system by aligning predictions across consecutive temporal windows. We observe that simple similarity transformation ($\mathrm{Sim}(3)$) alignment fails due to layer depth misalignment: monocular scale ambiguity causes relative depth scales of different scene layers to vary inconsistently between windows. To address this, we introduce layer-wise scale alignment, which segments depth predictions into discrete layers, computes per-layer scale factors, and propagates them across both adjacent windows and timestamps. Extensive experiments show that LASER achieves state-of-the-art performance on camera pose estimation and point map reconstruction %quality with offline models while operating at 14 FPS with 6 GB peak memory on a RTX A6000 GPU, enabling practical deployment for kilometer-scale streaming videos. Project website: $\href{https://neu-vi.github.io/LASER/}{\texttt{https://neu-vi.github.io/LASER/}}$
URLs: https://neu-vi.github.io/LASER/, https://neu-vi.github.io/LASER/
Authors: Lu Ling, Yunhao Ge, Yichen Sheng, Aniket Bera
Abstract: Generalization remains the central challenge for interactive 3D scene generation. Existing learning-based approaches ground spatial understanding in limited scene dataset, restricting generalization to new layouts. We instead reprogram a pre-trained 3D instance generator to act as a scene level learner, replacing dataset-bounded supervision with model-centric spatial supervision. This reprogramming unlocks the generator transferable spatial knowledge, enabling generalization to unseen layouts and novel object compositions. Remarkably, spatial reasoning still emerges even when the training scenes are randomly composed objects. This demonstrates that the generator's transferable scene prior provides a rich learning signal for inferring proximity, support, and symmetry from purely geometric cues. Replacing widely used canonical space, we instantiate this insight with a view-centric formulation of the scene space, yielding a fully feed-forward, generalizable scene generator that learns spatial relations directly from the instance model. Quantitative and qualitative results show that a 3D instance generator is an implicit spatial learner and reasoner, pointing toward foundation models for interactive 3D scene understanding and generation. Project page: https://luling06.github.io/I-Scene-project/
Authors: Daniel Zoran, Nikhil Parthasarathy, Yi Yang, Drew A Hudson, Joao Carreira, Andrew Zisserman
Abstract: We present Recurrent Video Masked-Autoencoders (RVM): a novel video representation learning approach that uses a transformer-based recurrent neural network to aggregate dense image features over time, effectively capturing the spatio-temporal structure of natural video data. RVM learns via an asymmetric masked prediction task requiring only a standard pixel reconstruction objective. This design yields a highly efficient ``generalist'' encoder: RVM achieves competitive performance with state-of-the-art video models (e.g. VideoMAE, V-JEPA) on video-level tasks like action recognition and point/object tracking, while also performing favorably against image models (e.g. DINOv2) on tasks that test geometric and dense spatial understanding. Notably, RVM achieves strong performance in the small-model regime without requiring knowledge distillation, exhibiting up to 30x greater parameter efficiency than competing video masked autoencoders. Moreover, we demonstrate that RVM's recurrent nature allows for stable feature propagation over long temporal horizons with linear computational cost, overcoming some of the limitations of standard spatio-temporal attention-based architectures. Finally, we use qualitative visualizations to highlight that RVM learns rich representations of scene semantics, structure, and motion.
Authors: Jingfeng Yao, Yuda Song, Yucong Zhou, Xinggang Wang
Abstract: The quality of the latent space in visual tokenizers (e.g., VAEs) is crucial for modern generative models. However, the standard reconstruction-based training paradigm produces a latent space that is biased towards low-level information, leading to a foundation flaw: better pixel-level accuracy does not lead to higher-quality generation. This implies that pouring extensive compute into visual tokenizer pre-training translates poorly to improved performance in generation. We identify this as the ``pre-training scaling problem`` and suggest a necessary shift: to be effective for generation, a latent space must concisely represent high-level semantics. We present VTP, a unified visual tokenizer pre-training framework, pioneering the joint optimization of image-text contrastive, self-supervised, and reconstruction losses. Our large-scale study reveals two principal findings: (1) understanding is a key driver of generation, and (2) much better scaling properties, where generative performance scales effectively with compute, parameters, and data allocated to the pretraining of the visual tokenizer. After large-scale pre-training, our tokenizer delivers a competitive profile (78.2 zero-shot accuracy and 0.36 rFID on ImageNet) and 4.1 times faster convergence on generation compared to advanced distillation methods. More importantly, it scales effectively: without modifying standard DiT training specs, solely investing more FLOPS in pretraining VTP achieves 65.8\% FID improvement in downstream generation, while conventional autoencoder stagnates very early at 1/10 FLOPS. Our pre-trained models are available at https://github.com/MiniMax-AI/VTP.
Authors: Yuanwen Yue, Damien Robert, Jianyuan Wang, Sunghwan Hong, Jan Dirk Wegner, Christian Rupprecht, Konrad Schindler
Abstract: Modern neural architectures for 3D point cloud processing contain both convolutional layers and attention blocks, but the best way to assemble them remains unclear. We analyse the role of different computational blocks in 3D point cloud networks and find an intuitive behaviour: convolution is adequate to extract low-level geometry at high-resolution in early layers, where attention is expensive without bringing any benefits; attention captures high-level semantics and context in low-resolution, deep layers more efficiently. Guided by this design principle, we propose a new, improved 3D point cloud backbone that employs convolutions in early stages and switches to attention for deeper layers. To avoid the loss of spatial layout information when discarding redundant convolution layers, we introduce a novel, training-free 3D positional encoding, PointROPE. The resulting LitePT model has $3.6\times$ fewer parameters, runs $2\times$ faster, and uses $2\times$ less memory than the state-of-the-art Point Transformer V3, but nonetheless matches or even outperforms it on a range of tasks and datasets. Code and models are available at: https://github.com/prs-eth/LitePT.
Authors: Susung Hong, Chongjian Ge, Zhifei Zhang, Jui-Hsien Wang
Abstract: Video diffusion models have revolutionized generative video synthesis, but they are imprecise, slow, and can be opaque during generation -- keeping users in the dark for a prolonged period. In this work, we propose DiffusionBrowser, a model-agnostic, lightweight decoder framework that allows users to interactively generate previews at any point (timestep or transformer block) during the denoising process. Our model can generate multi-modal preview representations that include RGB and scene intrinsics at more than 4$\times$ real-time speed (less than 1 second for a 4-second video) that convey consistent appearance and motion to the final video. With the trained decoder, we show that it is possible to interactively guide the generation at intermediate noise steps via stochasticity reinjection and modal steering, unlocking a new control capability. Moreover, we systematically probe the model using the learned decoders, revealing how scene, object, and other details are composed and assembled during the otherwise black-box denoising process.
Authors: Zheng Li, Peng Zhang, Shixiao Liang, Hang Zhou, Chengyuan Ma, Handong Yao, Qianwen Li, Xiaopeng Li
Abstract: Understanding how Advanced Driver-Assistance Systems (ADAS) interact with Traffic Control Devices (TCDs) is critical for assessing their influence on traffic operations, yet this interaction has received little focused empirical study. This paper presents a field dataset and behavioral analysis of Tesla's Traffic Light and Stop Sign Control (TLSSC), a mature ADAS that perceives traffic lights and stop signs. We design and execute experiments across varied speed limits and TCD types, collecting synchronized high-resolution vehicle trajectory data and driver-perspective video. From these data, we develop a taxonomy of TLSSC-TCD interaction behaviors (i.e., stopping, accelerating, and car following) and calibrate the Full Velocity Difference Model (FVDM) to quantitatively characterize each behavior mode. A novel empirical insight is the identification of a car-following threshold (~90 m). Calibration results reveal that stopping behavior is driven by strong responsiveness to both desired speed deviation and relative speed, whereas accelerating behavior is more conservative. Intersection car-following behavior exhibits smoother dynamics and tighter headways compared to standard car-following behaviors. The established dataset, behavior definitions, and model characterizations together provide a foundation for future simulation, safety evaluation, and design of ADAS-TCD interaction logic. Our dataset is available at GitHub.
Authors: Fengyi Xu, Jun Ma, Waishan Qiu, Cui Guo
Abstract: Crowdsourced street-view imagery from social media provides valuable real-time visual evidence of urban flooding and other crisis events, yet it often lacks reliable geographic metadata for emergency response. Existing Visual Place Recognition (VPR) models exhibit substantial performance degradation when applied to such imagery due to visual distortions and domain shifts inherent in cross-source scenarios. This paper presents VPR-AttLLM, a model-agnostic framework that integrates the semantic reasoning and geospatial knowledge of Large Language Models (LLMs) into established VPR pipelines through attention-guided descriptor enhancement. By leveraging LLMs to identify location-informative regions within the city context and suppress transient visual noise, VPR-AttLLM improves retrieval performance without requiring model retraining or additional data. Comprehensive evaluations are conducted on extended benchmarks including SF-XL enriched with real social-media flood images, synthetic flooding scenarios over established query sets and Mapillary photos, and a new HK-URBAN dataset capturing morphologically distinct cityscapes. Integrating VPR-AttLLM with three state-of-the-art VPR models-CosPlace, EigenPlaces, and SALAD-consistently improves recall performance, yielding relative gains typically between 1-3% and reaching up to 8% on the most challenging real flood imagery. Beyond measurable gains in retrieval accuracy, this study establishes a generalizable paradigm for LLM-guided multimodal fusion in visual retrieval systems. By embedding principles from urban perception theory into attention mechanisms, VPR-AttLLM bridges human-like spatial reasoning with modern VPR architectures. Its plug-and-play design, strong cross-source robustness, and interpretability highlight its potential for scalable urban monitoring and rapid geo-localization of crowdsourced crisis imagery.
Authors: Juan Palomeque-Gonzalez
Abstract: This technical note presents a reproducible workflow for converting a legacy archaeological image collection into a structured and segmentation ready dataset. The case study focuses on the Lower Palaeolithic hand axe and biface collection curated by the Archaeology Data Service (ADS), a dataset that provides thousands of standardised photographs but no mechanism for bulk download or automated processing. To address this, two open source tools were developed: a web scraping script that retrieves all record pages, extracts associated metadata, and downloads the available images while respecting ADS Terms of Use and ethical scraping guidelines; and an image processing pipeline that renames files using UUIDs, generates binary masks and bounding boxes through classical computer vision, and stores all derived information in a COCO compatible Json file enriched with archaeological metadata. The original images are not redistributed, and only derived products such as masks, outlines, and annotations are shared. Together, these components provide a lightweight and reusable approach for transforming web based archaeological image collections into machine learning friendly formats, facilitating downstream analysis and contributing to more reproducible research practices in digital archaeology.
Authors: Rosh Ho, Jian Zhang
Abstract: This paper presents ReGlove, a system that converts low-cost commercial pneumatic rehabilitation gloves into vision-guided assistive orthoses. Chronic upper-limb impairment affects millions worldwide, yet existing assistive technologies remain prohibitively expensive or rely on unreliable biological signals. Our platform integrates a wrist-mounted camera with an edge-computing inference engine (Raspberry Pi 5) to enable context-aware grasping without requiring reliable muscle signals. By adapting real-time YOLO-based computer vision models, the system achieves \SI{96.73}{\percent} grasp classification accuracy with sub-\SI{40.00}{\milli\second} end-to-end latency. Physical validation using standardized benchmarks shows \SI{82.71}{\percent} success on YCB object manipulation and reliable performance across \SI{27.00}{} Activities of Daily Living (ADL) tasks. With a total cost under \$\SI{250.00}{} and exclusively commercial components, ReGlove provides a technical foundation for accessible, vision-based upper-limb assistance that could benefit populations excluded from traditional EMG-controlled devices.
Authors: Milad Malekzadeh, Magdalena Biernacka, Elias Willberg, Jussi Torkko, Edyta {\L}aszkiewicz, Tuuli Toivonen
Abstract: Understanding greenspace attractiveness is essential for designing livable and inclusive urban environments, yet existing assessment approaches often overlook informal or transient spaces and remain too resource intensive to capture subjective perceptions at scale. This study examines the ability of multimodal large language models (MLLMs), ChatGPT GPT-4o, Claude 3.5 Haiku, and Gemini 2.0 Flash, to assess greenspace attractiveness similarly to humans using Google Street View imagery. We compared model outputs with responses from a geo-questionnaire of residents in Lodz, Poland, across both formal (for example, parks and managed greenspaces) and informal (for example, meadows and wastelands) greenspaces. Survey respondents and models indicated whether each greenspace was attractive or unattractive and provided up to three free text explanations. Analyses examined how often their attractiveness judgments aligned and compared their explanations after classifying them into shared reasoning categories. Results show high AI human agreement for attractive formal greenspaces and unattractive informal spaces, but low alignment for attractive informal and unattractive formal greenspaces. Models consistently emphasized aesthetic and design oriented features, underrepresenting safety, functional infrastructure, and locally embedded qualities valued by survey respondents. While these findings highlight the potential for scalable pre-assessment, they also underscore the need for human oversight and complementary participatory approaches. We conclude that MLLMs can support, but not replace, context sensitive greenspace evaluation in planning practice.
Authors: Haitao Lin, Peiyan Hu, Minsi Ren, Zhifeng Gao, Zhi-Ming Ma, Guolin ke, Tailin Wu, Stan Z. Li
Abstract: Recent advances in few-step diffusion models have demonstrated their efficiency and effectiveness by shortcutting the probabilistic paths of diffusion models, especially in training one-step diffusion models from scratch (a.k.a. shortcut models). However, their theoretical derivation and practical implementation are often closely coupled, which obscures the design space. To address this, we propose a common design framework for representative shortcut models. This framework provides theoretical justification for their validity and disentangles concrete component-level choices, thereby enabling systematic identification of improvements. With our proposed improvements, the resulting one-step model achieves a new state-of-the-art FID50k of 2.85 on ImageNet-256x256 under the classifier-free guidance setting. Remarkably, the model requires no pre-training, distillation, or curriculum learning. We believe our work lowers the barrier to component-level innovation in shortcut models and facilitates principled exploration of their design space.
Authors: Reuben R Shamir
Abstract: We implemented a Soft Decision Tree (SDT) and a Short-term Memory Soft Decision Tree (SM-SDT) using PyTorch. The methods were extensively tested on simulated and clinical datasets. The SDT was visualized to demonstrate the potential for its explainability. SDT, SM-SDT, and XGBoost demonstrated similar area under the curve (AUC) values. These methods were better than Random Forest, Logistic Regression, and Decision Tree. The results on clinical datasets suggest that, aside from a decision tree, all tested classification methods yield comparable results. The code and datasets are available online on GitHub: https://github.com/KI-Research-Institute/Soft-Decision-Tree
URLs: https://github.com/KI-Research-Institute/Soft-Decision-Tree
Authors: Mahmut S. Gokmen, Mitchell A. Klusty, Evan W. Damron, W. Vaiden Logan, Aaron D. Mullen, Caroline N. Leach, Emily B. Collier, Samuel E. Armstrong, V. K. Cody Bumgardner
Abstract: Self-supervised learning (SSL) leverages vast unannotated medical datasets, yet steep technical barriers limit adoption by clinical researchers. We introduce Vision Foundry, a code-free, HIPAA-compliant platform that democratizes pre-training, adaptation, and deployment of foundational vision models. The system integrates the DINO-MX framework, abstracting distributed infrastructure complexities while implementing specialized strategies like Magnification-Aware Distillation (MAD) and Parameter-Efficient Fine-Tuning (PEFT). We validate the platform across domains, including neuropathology segmentation, lung cellularity estimation, and coronary calcium scoring. Our experiments demonstrate that models trained via Vision Foundry significantly outperform generic baselines in segmentation fidelity and regression accuracy, while exhibiting robust zero-shot generalization across imaging protocols. By bridging the gap between advanced representation learning and practical application, Vision Foundry enables domain experts to develop state-of-the-art clinical AI tools with minimal annotation overhead, shifting focus from engineering optimization to clinical discovery.
Authors: Nimol Thuon, Jun Du
Abstract: Automated document layout analysis remains a major challenge for low-resource, non-Latin scripts. Khmer is a language spoken daily by over 17 million people in Cambodia, receiving little attention in the development of document AI tools. The lack of dedicated resources is particularly acute for business documents, which are critical for both public administration and private enterprise. To address this gap, we present \textbf{KH-FUNSD}, the first publicly available, hierarchically annotated dataset for Khmer form document understanding, including receipts, invoices, and quotations. Our annotation framework features a three-level design: (1) region detection that divides each document into core zones such as header, form field, and footer; (2) FUNSD-style annotation that distinguishes questions, answers, headers, and other key entities, together with their relationships; and (3) fine-grained classification that assigns specific semantic roles, such as field labels, values, headers, footers, and symbols. This multi-level approach supports both comprehensive layout analysis and precise information extraction. We benchmark several leading models, providing the first set of baseline results for Khmer business documents, and discuss the distinct challenges posed by non-Latin, low-resource scripts. The KH-FUNSD dataset and documentation will be available at URL.
Authors: Daniil Zverev, A. Sophia Koepke, Joao F. Henriques
Abstract: The use of synthetically generated data for training models is becoming a common practice. While generated data can augment the training data, repeated training on synthetic data raises concerns about distribution drift and degradation of performance due to contamination of the dataset. We investigate the consequences of this bootstrapping process through the lens of continual learning, drawing a connection to Generative Experience Replay (GER) methods. We present a statistical analysis showing that synthetic data introduces significant bias and variance into training objectives, weakening the reliability of maximum likelihood estimation. We provide empirical evidence showing that popular generative models collapse under repeated training with synthetic data. We quantify this degradation and show that state-of-the-art GER methods fail to maintain alignment in the latent space. Our findings raise critical concerns about the use of synthetic data in continual learning.
Authors: Wenqi Marshall Guo, Qingyun Qian, Khalad Hasan, Shan Du
Abstract: Over-aligning image generation models to a generalized aesthetic preference conflicts with user intent, particularly when ``anti-aesthetic" outputs are requested for artistic or critical purposes. This adherence prioritizes developer-centered values, compromising user autonomy and aesthetic pluralism. We test this bias by constructing a wide-spectrum aesthetics dataset and evaluating state-of-the-art generation and reward models. We find that aesthetic-aligned generation models frequently default to conventionally beautiful outputs, failing to respect instructions for low-quality or negative imagery. Crucially, reward models penalize anti-aesthetic images even when they perfectly match the explicit user prompt. We confirm this systemic bias through image-to-image editing and evaluation against real abstract artworks.
Authors: Iacopo Catalano, Eduardo Montijano, Javier Civera, Julio A. Placed, Jorge Pena-Queralta
Abstract: Autonomous navigation in dynamic environments requires spatial representations that capture both semantic structure and temporal evolution. 3D Scene Graphs (3DSGs) provide hierarchical multi-resolution abstractions that encode geometry and semantics, but existing extensions toward dynamics largely focus on individual objects or agents. In parallel, Maps of Dynamics (MoDs) model typical motion patterns and temporal regularities, yet are usually tied to grid-based discretizations that lack semantic awareness and do not scale well to large environments. In this paper we introduce Aion, a framework that embeds temporal flow dynamics directly within a hierarchical 3DSG, effectively incorporating the temporal dimension. Aion employs a graph-based sparse MoD representation to capture motion flows over arbitrary time intervals and attaches them to navigational nodes in the scene graph, yielding more interpretable and scalable predictions that improve planning and interaction in complex dynamic environments.
Authors: Nabeel Rehemtulla, Adam A. Miller, Mike Walmsley, Ved G. Shah, Theophile Jegou du Laz, Michael W. Coughlin, Argyro Sasli, Joshua Bloom, Christoffer Fremling, Matthew J. Graham, Steven L. Groom, David Hale, Ashish A. Mahabal, Daniel A. Perley, Josiah Purdum, Ben Rusholme, Jesper Sollerman, Mansi M. Kasliwal
Abstract: Modern wide-field time-domain surveys facilitate the study of transient, variable and moving phenomena by conducting image differencing and relaying alerts to their communities. Machine learning tools have been used on data from these surveys and their precursors for more than a decade, and convolutional neural networks (CNNs), which make predictions directly from input images, saw particularly broad adoption through the 2010s. Since then, continually rapid advances in computer vision have transformed the standard practices around using such models. It is now commonplace to use standardized architectures pre-trained on large corpora of everyday images (e.g., ImageNet). In contrast, time-domain astronomy studies still typically design custom CNN architectures and train them from scratch. Here, we explore the affects of adopting various pre-training regimens and standardized model architectures on the performance of alert classification. We find that the resulting models match or outperform a custom, specialized CNN like what is typically used for filtering alerts. Moreover, our results show that pre-training on galaxy images from Galaxy Zoo tends to yield better performance than pre-training on ImageNet or training from scratch. We observe that the design of standardized architectures are much better optimized than the custom CNN baseline, requiring significantly less time and memory for inference despite having more trainable parameters. On the eve of the Legacy Survey of Space and Time and other image-differencing surveys, these findings advocate for a paradigm shift in the creation of vision models for alerts, demonstrating that greater performance and efficiency, in time and in data, can be achieved by adopting the latest practices from the computer vision field.
Authors: Nolan Koblischke, Liam Parker, Francois Lanusse, Irina Espejo Morales, Jo Bovy, Shirley Ho
Abstract: Finding scientifically interesting phenomena through slow, manual labeling campaigns severely limits our ability to explore the billions of galaxy images produced by telescopes. In this work, we develop a pipeline to create a semantic search engine from completely unlabeled image data. Our method leverages Vision-Language Models (VLMs) to generate descriptions for galaxy images, then contrastively aligns a pre-trained multimodal astronomy foundation model with these embedded descriptions to produce searchable embeddings at scale. We find that current VLMs provide descriptions that are sufficiently informative to train a semantic search model that outperforms direct image similarity search. Our model, AION-Search, achieves state-of-the-art zero-shot performance on finding rare phenomena despite training on randomly selected images with no deliberate curation for rare cases. Furthermore, we introduce a VLM-based re-ranking method that nearly doubles the recall for our most challenging targets in the top-100 results. For the first time, AION-Search enables flexible semantic search scalable to 140 million galaxy images, enabling discovery from previously infeasible searches. More broadly, our work provides an approach for making large, unlabeled scientific image archives semantically searchable, expanding data exploration capabilities in fields from Earth observation to microscopy. The code, data, and app are publicly available at https://github.com/NolanKoblischke/AION-Search
Authors: Xiaoxuan Tang, Xinping Lei, Chaoran Zhu, Shiyun Chen, Ruibin Yuan, Yizhi Li, Changjae Oh, Ge Zhang, Wenhao Huang, Emmanouil Benetos, Yang Liu, Jiaheng Liu, Yinghao Ma
Abstract: Music-to-Video (M2V) generation for full-length songs faces significant challenges. Existing methods produce short, disjointed clips, failing to align visuals with musical structure, beats, or lyrics, and lack temporal consistency. We propose AutoMV, a multi-agent system that generates full music videos (MVs) directly from a song. AutoMV first applies music processing tools to extract musical attributes, such as structure, vocal tracks, and time-aligned lyrics, and constructs these features as contextual inputs for following agents. The screenwriter Agent and director Agent then use this information to design short script, define character profiles in a shared external bank, and specify camera instructions. Subsequently, these agents call the image generator for keyframes and different video generators for "story" or "singer" scenes. A Verifier Agent evaluates their output, enabling multi-agent collaboration to produce a coherent longform MV. To evaluate M2V generation, we further propose a benchmark with four high-level categories (Music Content, Technical, Post-production, Art) and twelve ine-grained criteria. This benchmark was applied to compare commercial products, AutoMV, and human-directed MVs with expert human raters: AutoMV outperforms current baselines significantly across all four categories, narrowing the gap to professional MVs. Finally, we investigate using large multimodal models as automatic MV judges; while promising, they still lag behind human expert, highlighting room for future work.
Authors: Eric J. Elias, Michael Esswein, Jonathan P. How, David W. Miller
Abstract: As the popularity of on-orbit operations grows, so does the need for precise navigation around unknown resident space objects (RSOs) such as other spacecraft, orbital debris, and asteroids. The use of Simultaneous Localization and Mapping (SLAM) algorithms is often studied as a method to map out the surface of an RSO and find the inspector's relative pose using a lidar or conventional camera. However, conventional cameras struggle during eclipse or shadowed periods, and lidar, though robust to lighting conditions, tends to be heavier, bulkier, and more power-intensive. Thermal-infrared cameras can track the target RSO throughout difficult illumination conditions without these limitations. While useful, thermal-infrared imagery lacks the resolution and feature-richness of visible cameras. In this work, images of a target satellite in low Earth orbit are photo-realistically simulated in both visible and thermal-infrared bands. Pixel-level fusion methods are used to create visible/thermal-infrared composites that leverage the best aspects of each camera. Navigation errors from a monocular SLAM algorithm are compared between visible, thermal-infrared, and fused imagery in various lighting and trajectories. Fused imagery yields substantially improved navigation performance over visible-only and thermal-only methods.
Authors: Aujasvit Datta, Jiayun Wang, Asad Aali, Armeet Singh Jatyani, Anima Anandkumar
Abstract: Sparse-view Computed Tomography (CT) reconstructs images from a limited number of X-ray projections to reduce radiation and scanning time, which makes reconstruction an ill-posed inverse problem. Deep learning methods achieve high-fidelity reconstructions but often overfit to a fixed acquisition setup, failing to generalize across sampling rates and image resolutions. For example, convolutional neural networks (CNNs) use the same learned kernels across resolutions, leading to artifacts when data resolution changes. We propose Computed Tomography neural Operator (CTO), a unified CT reconstruction framework that extends to continuous function space, enabling generalization (without retraining) across sampling rates and image resolutions. CTO operates jointly in the sinogram and image domains through rotation-equivariant Discrete-Continuous convolutions parametrized in the function space, making it inherently resolution- and sampling-agnostic. Empirically, CTO enables consistent multi-sampling-rate and cross-resolution performance, with on average >4dB PSNR gain over CNNs. Compared to state-of-the-art diffusion methods, CTO is 500$\times$ faster in inference time with on average 3dB gain. Empirical results also validate our design choices behind CTO's sinogram-space operator learning and rotation-equivariant convolution. Overall, CTO outperforms state-of-the-art baselines across sampling rates and resolutions, offering a scalable and generalizable solution that makes automated CT reconstruction more practical for deployment.
Authors: Donghyuk Kim, Sejeong Yang, Wonjin Shin, Joo-Young Kim
Abstract: Streaming video large language models (LLMs) are increasingly used for real-time multimodal tasks such as video captioning, question answering, conversational agents, and augmented reality. However, these models face fundamental memory and computational challenges because their key-value (KV) caches grow substantially with continuous streaming video input. This process requires an iterative prefill stage, which is a unique feature of streaming video LLMs. Due to its iterative prefill stage, it suffers from significant limitations, including extensive computation, substantial data transfer, and degradation in accuracy. Crucially, this issue is exacerbated for edge deployment, which is the primary target for these models. In this work, we propose V-Rex, the first software-hardware co-designed accelerator that comprehensively addresses both algorithmic and hardware bottlenecks in streaming video LLM inference. At its core, V-Rex introduces ReSV, a training-free dynamic KV cache retrieval algorithm. ReSV exploits temporal and spatial similarity-based token clustering to reduce excessive KV cache memory across video frames. To fully realize these algorithmic benefits, V-Rex offers a compact, low-latency hardware accelerator with a dynamic KV cache retrieval engine (DRE), featuring bit-level and early-exit based computing units. V-Rex achieves unprecedented real-time of 3.9-8.3 FPS and energy-efficient streaming video LLM inference on edge deployment with negligible accuracy loss. While DRE only accounts for 2.2% power and 2.0% area, the system delivers 1.9-19.7x speedup and 3.1-18.5x energy efficiency improvements over AGX Orin GPU. This work is the first to comprehensively tackle KV cache retrieval across algorithms and hardware, enabling real-time streaming video LLM inference on resource-constrained edge devices.
Authors: Shuyang Xie, Jie Zhou, Jun Wang, Renjing Xu
Abstract: Computer-generated holography (CGH) presents a transformative solution for near-eye displays in augmented and virtual reality. Recent advances in deep learning have greatly improved CGH in reconstructed quality and computational efficiency. However, deploying neural CGH pipelines directly on compact, eyeglass-style devices is hindered by stringent constraints on computation and energy consumption, while cloud offloading followed by transmission with natural image codecs often distorts phase information and requires high bandwidth to maintain reconstruction quality. Neural compression methods can reduce bandwidth but impose heavy neural decoders at the edge, increasing inference latency and hardware demand. In this work, we introduce JPEG-Inspired Cloud-Edge Holography, an efficient pipeline designed around a learnable transform codec that retains the block-structured and hardware-friendly nature of JPEG. Our system shifts all heavy neural processing to the cloud, while the edge device performs only lightweight decoding without any neural inference. To further improve throughput, we implement custom CUDA kernels for entropy coding on both cloud and edge. This design achieves a peak signal-to-noise ratio of 32.15 dB at $<$ 2 bits per pixel with decode latency as low as 4.2 ms. Both numerical simulations and optical experiments confirm the high reconstruction quality of the holograms. By aligning CGH with a codec that preserves JPEG's structural efficiency while extending it with learnable components, our framework enables low-latency, bandwidth-efficient hologram streaming on resource-constrained wearable devices-using only simple block-based decoding readily supported by modern system-on-chips, without requiring neural decoders or specialized hardware.
Authors: Gelesh G Omathil, Sreeja CS
Abstract: Deep neural networks possess strong representational capacity yet remain vulnerable to overfitting, primarily because neurons tend to co-adapt in ways that, while capturing complex and fine-grained feature interactions, also reinforce spurious and non-generalizable patterns that inflate training performance but reduce reliability on unseen data. Noise-based regularizers such as Dropout and DropConnect address this issue by injecting stochastic perturbations during training, but the noise they apply is typically uniform across a layer or across a batch of samples, which can suppress both harmful and beneficial co-adaptation. This work introduces PerNodeDrop, a lightweight stochastic regularization method. It applies per-sample, per-node perturbations to break the uniformity of the noise injected by existing techniques, thereby allowing each node to experience input-specific variability. Hence, PerNodeDrop preserves useful co-adaptation while applying regularization. This narrows the gap between training and validation performance and improves reliability on unseen data, as evident from the experiments. Although superficially similar to DropConnect, PerNodeDrop operates at the sample level. It drops weights at the sample level, not the batch level. An expected-loss analysis formalizes how its perturbations attenuate excessive co-adaptation while retaining predictive interactions. Empirical evaluations on vision, text, and audio benchmarks indicate improved generalization relative to the standard noise-based regularizer.
Authors: Yeray Cordero, Paula Garc\'ia-Molina, Fernando Vilari\~no
Abstract: Implicit neural representations (INRs) have become a powerful paradigm for continuous signal modeling and 3D scene reconstruction, yet classical networks suffer from a well-known spectral bias that limits their ability to capture high-frequency details. Quantum Implicit Representation Networks (QIREN) mitigate this limitation by employing parameterized quantum circuits with inherent Fourier structures, enabling compact and expressive frequency modeling beyond classical MLPs. In this paper, we present Quantum Neural Radiance Fields (Q-NeRF), the first hybrid quantum-classical framework for neural radiance field rendering. Q-NeRF integrates QIREN modules into the Nerfacto backbone, preserving its efficient sampling, pose refinement, and volumetric rendering strategies while replacing selected density and radiance prediction components with quantum-enhanced counterparts. We systematically evaluate three hybrid configurations on standard multi-view indoor datasets, comparing them to classical baselines using PSNR, SSIM, and LPIPS metrics. Results show that hybrid quantum-classical models achieve competitive reconstruction quality under limited computational resources, with quantum modules particularly effective in representing fine-scale, view-dependent appearance. Although current implementations rely on quantum circuit simulators constrained to few-qubit regimes, the results highlight the potential of quantum encodings to alleviate spectral bias in implicit representations. Q-NeRF provides a foundational step toward scalable quantum-enabled 3D scene reconstruction and a baseline for future quantum neural rendering research.
Authors: Yongcan Yu, Lingxiao He, Shuo Lu, Lijun Sheng, Yinuo Xu, Yanbo Wang, Kuangpu Guo, Jianjie Cheng, Meng Wang, Qianlong Xie, Xingxing Wang, Dapeng Hu, Jian Liang
Abstract: Recent advances in vision-language models (VLMs) reasoning have been largely attributed to the rise of reinforcement Learning (RL), which has shifted the community's focus away from the supervised fine-tuning (SFT) paradigm. Many studies suggest that introducing the SFT stage not only fails to improve reasoning ability but may also negatively impact model training. In this study, we revisit this RL-centric belief through a systematic and controlled comparison of SFT and RL on VLM Reasoning. Using identical data sources, we find that the relative effectiveness of SFT and RL is conditional and strongly influenced by model capacity, data scale, and data distribution. Contrary to common assumptions, our findings show that SFT plays a crucial role across several scenarios: (1) Effectiveness for weaker models. SFT more reliably elicits reasoning capabilities in smaller or weaker VLMs. (2) Data efficiency. SFT with only 2K achieves comparable or better reasoning performance to RL with 20K. (3) Cross-modal transferability. SFT demonstrates stronger generalization across modalities. Moreover, we identify a pervasive issue of deceptive rewards, where higher rewards fail to correlate with better reasoning accuracy in RL. These results challenge the prevailing "RL over SFT" narrative. They highlight that the role of SFT may have been underestimated and support a more balanced post-training pipeline in which SFT and RL function as complementary components.
Authors: Anthony Mudet, Souhail Bakkali
Abstract: Large-scale digitization initiatives have unlocked massive collections of historical newspapers, yet effective computational access remains hindered by OCR corruption, multilingual orthographic variation, and temporal language drift. We develop and evaluate a multilingual Retrieval-Augmented Generation pipeline specifically designed for question answering on noisy historical documents. Our approach integrates: (i) semantic query expansion and multi-query fusion using Reciprocal Rank Fusion to improve retrieval robustness against vocabulary mismatch; (ii) a carefully engineered generation prompt that enforces strict grounding in retrieved evidence and explicit abstention when evidence is insufficient; and (iii) a modular architecture enabling systematic component evaluation. We conduct comprehensive ablation studies on Named Entity Recognition and embedding model selection, demonstrating the importance of syntactic coherence in entity extraction and balanced performance-efficiency trade-offs in dense retrieval. Our end-to-end evaluation framework shows that the pipeline generates faithful answers for well-supported queries while correctly abstaining from unanswerable questions. The hybrid retrieval strategy improves recall stability, particularly benefiting from RRF's ability to smooth performance variance across query formulations. We release our code and configurations at https://anonymous.4open.science/r/RAGs-C5AE/, providing a reproducible foundation for robust historical document question answering.
Authors: Incheol Baek, Hyungbin Kim, Minseo Kim, Yon Dohn Chung
Abstract: Federated Learning (FL) enables collaborative training across multiple clients while preserving data privacy, yet it struggles with data heterogeneity, where clients' data are not distributed independently and identically (non-IID). This causes local drift, hindering global model convergence. To address this, we introduce Federated Learning with Feedback Alignment (FLFA), a novel framework that integrates feedback alignment into FL. FLFA uses the global model's weights as a shared feedback matrix during local training's backward pass, aligning local updates with the global model efficiently. This approach mitigates local drift with minimal additional computational cost and no extra communication overhead. Our theoretical analysis supports FLFA's design by showing how it alleviates local drift and demonstrates robust convergence for both local and global models. Empirical evaluations, including accuracy comparisons and measurements of local drift, further illustrate that FLFA can enhance other FL methods demonstrating its effectiveness.
Authors: Jianghan Chao, Jianzhang Gao, Wenhui Tan, Yuchong Sun, Ruihua Song, Liyun Ru
Abstract: Understanding videos inherently requires reasoning over both visual and auditory information. To properly evaluate Omni-Large Language Models (Omni-LLMs), which are capable of processing multi-modal information including vision and audio, an effective benchmark must comprehensively cover three key aspects: (1) multi-modal dependency (i.e., questions that cannot be answered using vision or audio alone), (2) diverse audio information types (e.g., speech, sound events), and (3) varying scene spans. However, existing datasets fall short in one or more of these dimensions, limiting strict and comprehensive evaluation. To address this gap, we introduce JointAVBench, a novel benchmark with strict audio-video correlation, spanning five cognitive dimensions, four audio information types (speech, sound events, music, vocal traits), and three scene spans (single-, cross-, and full-scene). Given the high cost of manual annotation, we propose an automated pipeline that leverages state-of-the-art vision-LLMs, audio-LLMs, and general-purpose LLMs to synthesize questions and answers that strictly require joint audio-visual understanding. We evaluate leading vision-only, audio-only, and Omni-LLMs on our dataset. Results show that even the best-performing Omni-LLM achieves an average accuracy of only 62.6\%, outperforming uni-modal baselines but revealing substantial room for improvement, especially in cross-scene reasoning.
Authors: Mohammad Mahdi Razmjoo, Mohammad Mahdi Sharifian, Saeed Bagheri Shouraki
Abstract: Despite their remarkable performance, deep neural networks exhibit a critical vulnerability: small, often imperceptible, adversarial perturbations can lead to drastically altered model predictions. Given the stringent reliability demands of applications such as medical diagnosis and autonomous driving, robust detection of such adversarial attacks is paramount. In this paper, we investigate the geometric properties of a model's input loss landscape. We analyze the Intrinsic Dimensionality (ID) of the model's gradient parameters, which quantifies the minimal number of coordinates required to describe the data points on their underlying manifold. We reveal a distinct and consistent difference in the ID for natural and adversarial data, which forms the basis of our proposed detection method. We validate our approach across two distinct operational scenarios. First, in a batch-wise context for identifying malicious data groups, our method demonstrates high efficacy on datasets like MNIST and SVHN. Second, in the critical individual-sample setting, we establish new state-of-the-art results on challenging benchmarks such as CIFAR-10 and MS COCO. Our detector significantly surpasses existing methods against a wide array of attacks, including CW and AutoAttack, achieving detection rates consistently above 92\% on CIFAR-10. The results underscore the robustness of our geometric approach, highlighting that intrinsic dimensionality is a powerful fingerprint for adversarial detection across diverse datasets and attack strategies.
Authors: Sebastien Tchitchek, Mohamed Kissi, Julien Tierny
Abstract: We introduce the Continuous Edit Distance (CED), a geodesic and elastic distance for time-varying persistence diagrams (TVPDs). The CED extends edit-distance ideas to TVPDs by combining local substitution costs with penalized deletions/insertions, controlled by two parameters: \(\alpha\) (trade-off between temporal misalignment and diagram discrepancy) and \(\beta\) (gap penalty). We also provide an explicit construction of CED-geodesics. Building on these ingredients, we present two practical barycenter solvers, one stochastic and one greedy, that monotonically decrease the CED Frechet energy. Empirically, the CED is robust to additive perturbations (both temporal and spatial), recovers temporal shifts, and supports temporal pattern search. On real-life datasets, the CED achieves clustering performance comparable or better than standard elastic dissimilarities, while our clustering based on CED-barycenters yields superior classification results. Overall, the CED equips TVPD analysis with a principled distance, interpretable geodesics, and practical barycenters, enabling alignment, comparison, averaging, and clustering directly in the space of TVPDs. A C++ implementation is provided for reproducibility at the following address https://github.com/sebastien-tchitchek/ContinuousEditDistance.
URLs: https://github.com/sebastien-tchitchek/ContinuousEditDistance.
Authors: Anja Sheppard, Parker Ewen, Joey Wilson, Advaith V. Sethuraman, Benard Adewole, Anran Li, Yuzhen Chen, Ram Vasudevan, Katherine A. Skinner
Abstract: This paper introduces SLIM-VDB, a new lightweight semantic mapping system with probabilistic semantic fusion for closed-set or open-set dictionaries. Advances in data structures from the computer graphics community, such as OpenVDB, have demonstrated significantly improved computational and memory efficiency in volumetric scene representation. Although OpenVDB has been used for geometric mapping in robotics applications, semantic mapping for scene understanding with OpenVDB remains unexplored. In addition, existing semantic mapping systems lack support for integrating both fixed-category and open-language label predictions within a single framework. In this paper, we propose a novel 3D semantic mapping system that leverages the OpenVDB data structure and integrates a unified Bayesian update framework for both closed- and open-set semantic fusion. Our proposed framework, SLIM-VDB, achieves significant reduction in both memory and integration times compared to current state-of-the-art semantic mapping approaches, while maintaining comparable mapping accuracy. An open-source C++ codebase with a Python interface is available at https://github.com/umfieldrobotics/slim-vdb.
Authors: Krishna Srikar Durbha, Hassene Tmar, Ping-Hao Wu, Ioannis Katsavounidis, Alan C. Bovik
Abstract: Over the past few years, per-title and per-shot video encoding techniques have demonstrated significant gains as compared to conventional techniques such as constant CRF encoding and the fixed bitrate ladder. These techniques have demonstrated that constructing content-gnostic per-shot bitrate ladders can provide significant bitrate gains and improved Quality of Experience (QoE) for viewers under various network conditions. However, constructing a convex hull for every video incurs a significant computational overhead. Recently, machine learning-based bitrate ladder construction techniques have emerged as a substitute for convex hull construction. These methods operate by extracting features from source videos to train machine learning (ML) models to construct content-adaptive bitrate ladders. Here, we present a new ML-based bitrate ladder construction technique that accurately predicts the VMAF scores of compressed videos, by analyzing the compression procedure and by making perceptually relevant measurements on the source videos prior to compression. We evaluate the performance of our proposed framework against leading prior methods on a large corpus of videos. Since training ML models on every encoder setting is time-consuming, we also investigate how per-shot bitrate ladders perform under different encoding settings. We evaluate the performance of all models against the fixed bitrate ladder and the best possible convex hull constructed using exhaustive encoding with Bjontegaard-delta metrics.
Authors: Jiayin Lu, Ying Jiang, Yin Yang, Chenfanfu Jiang
Abstract: We present VoroLight, a differentiable framework for 3D shape reconstruction based on Voronoi meshing. Our approach generates smooth, watertight surfaces and topologically consistent volumetric meshes directly from diverse inputs, including images, implicit shape level-set fields, point clouds and meshes. VoroLight operates in three stages: it first initializes a surface using a differentiable Voronoi formulation, then refines surface quality through a polygon-face sphere training stage, and finally reuses the differentiable Voronoi formulation for volumetric optimization with additional interior generator points. Project page: https://jiayinlu19960224.github.io/vorolight/
Authors: Amin Jalal Aghdasian, Farzaneh Abdollahi, Ali Kamali Iglie
Abstract: This paper proposes two new algorithms for the lane keeping system (LKS) in autonomous vehicles (AVs) operating under snowy road conditions. These algorithms use deep reinforcement learning (DRL) to handle uncertainties and slippage. They include Action-Robust Recurrent Deep Deterministic Policy Gradient (AR-RDPG) and end-to-end Action-Robust convolutional neural network Attention Deterministic Policy Gradient (AR-CADPG), two action-robust approaches for decision-making. In the AR-RDPG method, within the perception layer, camera images are first denoised using multi-scale neural networks. Then, the centerline coefficients are extracted by a pre-trained deep convolutional neural network (DCNN). These coefficients, concatenated with the driving characteristics, are used as input to the control layer. The AR-CADPG method presents an end-to-end approach in which a convolutional neural network (CNN) and an attention mechanism are integrated within a DRL framework. Both methods are first trained in the CARLA simulator and validated under various snowy scenarios. Real-world experiments on a Jetson Nano-based autonomous vehicle confirm the feasibility and stability of the learned policies. Among the two models, the AR-CADPG approach demonstrates superior path-tracking accuracy and robustness, highlighting the effectiveness of combining temporal memory, adversarial resilience, and attention mechanisms in AVs.
Authors: Xin Guo, Yifan Zhao, Jia Li
Abstract: Generating 3D-based body movements from speech shows great potential in extensive downstream applications, while it still suffers challenges in imitating realistic human movements. Predominant research efforts focus on end-to-end generation schemes to generate co-speech gestures, spanning GANs, VQ-VAE, and recent diffusion models. As an ill-posed problem, in this paper, we argue that these prevailing learning schemes fail to model crucial inter- and intra-correlations across different motion units, i.e. head, body, and hands, thus leading to unnatural movements and poor coordination. To delve into these intrinsic correlations, we propose a unified Hierarchical Implicit Periodicity (HIP) learning approach for audio-inspired 3D gesture generation. Different from predominant research, our approach models this multi-modal implicit relationship by two explicit technique insights: i) To disentangle the complicated gesture movements, we first explore the gesture motion phase manifolds with periodic autoencoders to imitate human natures from realistic distributions while incorporating non-period ones from current latent states for instance-level diversities. ii) To model the hierarchical relationship of face motions, body gestures, and hand movements, driving the animation with cascaded guidance during learning. We exhibit our proposed approach on 3D avatars and extensive experiments show our method outperforms the state-of-the-art co-speech gesture generation methods by both quantitative and qualitative evaluations. Code and models will be publicly available.
Authors: Hyunki Seong, Jeong-Kyun Lee, Heesoo Myeong, Yongho Shin, Hyun-Mook Cho, Duck Hoon Kim, Pranav Desai, Monu Surana
Abstract: Learning interactive motion behaviors among multiple agents is a core challenge in autonomous driving. While imitation learning models generate realistic trajectories, they often inherit biases from datasets dominated by safe demonstrations, limiting robustness in safety-critical cases. Moreover, most studies rely on open-loop evaluation, overlooking compounding errors in closed-loop execution. We address these limitations with two complementary strategies. First, we propose Group Relative Behavior Optimization (GRBO), a reinforcement learning post-training method that fine-tunes pretrained behavior models via group relative advantage maximization with human regularization. Using only 10% of the training dataset, GRBO improves safety performance by over 40% while preserving behavioral realism. Second, we introduce Warm-K, a warm-started Top-K sampling strategy that balances consistency and diversity in motion selection. Our Warm-K method-based test-time scaling enhances behavioral consistency and reactivity at test time without retraining, mitigating covariate shift and reducing performance discrepancies. Demo videos are available in the supplementary material.
Authors: Youssef Megahed, Inok Lee, Robin Ducharme, Kevin Dick, Adrian D. C. Chan, Steven Hawken, Mark C. Walker
Abstract: Prenatal ultrasound is the cornerstone for detecting congenital anomalies of the kidneys and urinary tract, but diagnosis is limited by operator dependence and suboptimal imaging conditions. We sought to assess the performance of a self-supervised ultrasound foundation model for automated fetal renal anomaly classification using a curated dataset of 969 two-dimensional ultrasound images. A pretrained Ultrasound Self-Supervised Foundation Model with Masked Autoencoding (USF-MAE) was fine-tuned for binary and multi-class classification of normal kidneys, urinary tract dilation, and multicystic dysplastic kidney. Models were compared with a DenseNet-169 convolutional baseline using cross-validation and an independent test set. USF-MAE consistently improved upon the baseline across all evaluation metrics in both binary and multi-class settings. USF-MAE achieved an improvement of about 1.87% (AUC) and 7.8% (F1-score) on the validation set, 2.32% (AUC) and 4.33% (F1-score) on the independent holdout test set. The largest gains were observed in the multi-class setting, where the improvement in AUC was 16.28% and 46.15% in F1-score. To facilitate model interpretability, Score-CAM visualizations were adapted for a transformer architecture and show that model predictions were informed by known, clinically relevant renal structures, including the renal pelvis in urinary tract dilation and cystic regions in multicystic dysplastic kidney. These results show that ultrasound-specific self-supervised learning can generate a useful representation as a foundation for downstream diagnostic tasks. The proposed framework offers a robust, interpretable approach to support the prenatal detection of renal anomalies and demonstrates the promise of foundation models in obstetric imaging.
Authors: Haoyu Ren, Kay Koehle, Kirill Dorofeev, Darko Anicic
Abstract: In modern manufacturing, Visual Anomaly Detection (VAD) is essential for automated inspection and consistent product quality. Yet, increasingly dynamic and flexible production environments introduce key challenges: First, frequent product changes in small-batch and on-demand manufacturing require rapid model updates. Second, legacy edge hardware lacks the resources to train and run large AI models. Finally, both anomalous and normal training data are often scarce, particularly for newly introduced product variations. We investigate on-device continual learning for unsupervised VAD with localization, extending the PatchCore to incorporate online learning for real-world industrial scenarios. The proposed method leverages a lightweight feature extractor and an incremental coreset update mechanism based on k-center selection, enabling rapid, memory-efficient adaptation from limited data while eliminating costly cloud retraining. Evaluations on an industrial use case are conducted using a testbed designed to emulate flexible production with frequent variant changes in a controlled environment. Our method achieves a 12% AUROC improvement over the baseline, an 80% reduction in memory usage, and faster training compared to batch retraining. These results confirm that our method delivers accurate, resource-efficient, and adaptive VAD suitable for dynamic and smart manufacturing.
Authors: Fu-Yun Wang, Hao Zhou, Liangzhe Yuan, Sanghyun Woo, Boqing Gong, Bohyung Han, Ming-Hsuan Yang, Han Zhang, Yukun Zhu, Ting Liu, Long Zhao
Abstract: The slow inference process of image diffusion models significantly degrades interactive user experiences. To address this, we introduce Diffusion Preview, a novel paradigm employing rapid, low-step sampling to generate preliminary outputs for user evaluation, deferring full-step refinement until the preview is deemed satisfactory. Existing acceleration methods, including training-free solvers and post-training distillation, struggle to deliver high-quality previews or ensure consistency between previews and final outputs. We propose ConsistencySolver derived from general linear multistep methods, a lightweight, trainable high-order solver optimized via Reinforcement Learning, that enhances preview quality and consistency. Experimental results demonstrate that ConsistencySolver significantly improves generation quality and consistency in low-step scenarios, making it ideal for efficient preview-and-refine workflows. Notably, it achieves FID scores on-par with Multistep DPM-Solver using 47% fewer steps, while outperforming distillation baselines. Furthermore, user studies indicate our approach reduces overall user interaction time by nearly 50% while maintaining generation quality. Code is available at https://github.com/G-U-N/consolver.
Authors: Gabriel Vitorino de Andrade, Saulo Roberto dos Santos, Itallo Patrick Castro Alves da Silva, Emanuel Adler Medeiros Pereira, Erick de Andrade Barboza
Abstract: The validation and verification of artificial intelligence (AI) models through robustness assessment are essential to guarantee the reliable performance of intelligent systems facing real-world challenges, such as image corruptions including noise, blurring, and weather variations. Despite the global importance of mango (Mangifera indica L.), there is a lack of studies on the robustness of models for the diagnosis of disease in its leaves. This paper proposes a methodology to evaluate convolutional neural networks (CNNs) under adverse conditions. We adapted the MangoLeafDB dataset, generating MangoLeafDB-C with 19 types of artificial corruptions at five severity levels. We conducted a benchmark comparing five architectures: ResNet-50, ResNet-101, VGG-16, Xception, and LCNN (the latter being a lightweight architecture designed specifically for mango leaf diagnosis). The metrics include the F1 score, the corruption error (CE) and the relative mean corruption error (relative mCE). The results show that LCNN outperformed complex models in corruptions that can be present in real-world scenarios such as Defocus Blur, Motion Blur, while also achieving the lowest mCE. Modern architectures (e.g., ResNet-101) exhibited significant performance degradation in corrupted scenarios, despite their high accuracy under ideal conditions. These findings suggest that lightweight and specialized models may be more suitable for real-world applications in edge devices, where robustness and efficiency are critical. The study highlights the need to incorporate robustness assessments in the development of intelligent systems for agriculture, particularly in regions with technological limitations.
Authors: Raktim Gautam Goswami, Amir Bar, David Fan, Tsung-Yen Yang, Gaoyue Zhou, Prashanth Krishnamurthy, Michael Rabbat, Farshad Khorrami, Yann LeCun
Abstract: Dexterous manipulation is challenging because it requires understanding how subtle hand motion influences the environment through contact with objects. We introduce DexWM, a Dexterous Manipulation World Model that predicts the next latent state of the environment conditioned on past states and dexterous actions. To overcome the scarcity of dexterous manipulation datasets, DexWM is trained on over 900 hours of human and non-dexterous robot videos. To enable fine-grained dexterity, we find that predicting visual features alone is insufficient; therefore, we introduce an auxiliary hand consistency loss that enforces accurate hand configurations. DexWM outperforms prior world models conditioned on text, navigation, and full-body actions, achieving more accurate predictions of future states. DexWM also demonstrates strong zero-shot generalization to unseen manipulation skills when deployed on a Franka Panda arm equipped with an Allegro gripper, outperforming Diffusion Policy by over 50% on average in grasping, placing, and reaching tasks.
Authors: Enshen Zhou, Cheng Chi, Yibo Li, Jingkun An, Jiayuan Zhang, Shanyu Rong, Yi Han, Yuheng Ji, Mengzhen Liu, Pengwei Wang, Zhongyuan Wang, Lu Sheng, Shanghang Zhang
Abstract: Spatial tracing, as a fundamental embodied interaction ability for robots, is inherently challenging as it requires multi-step metric-grounded reasoning compounded with complex spatial referring and real-world metric measurement. However, existing methods struggle with this compositional task. To this end, we propose RoboTracer, a 3D-aware VLM that first achieves both 3D spatial referring and measuring via a universal spatial encoder and a regression-supervised decoder to enhance scale awareness during supervised fine-tuning (SFT). Moreover, RoboTracer advances multi-step metric-grounded reasoning via reinforcement fine-tuning (RFT) with metric-sensitive process rewards, supervising key intermediate perceptual cues to accurately generate spatial traces. To support SFT and RFT training, we introduce TraceSpatial, a large-scale dataset of 30M QA pairs, spanning outdoor/indoor/tabletop scenes and supporting complex reasoning processes (up to 9 steps). We further present TraceSpatial-Bench, a challenging benchmark filling the gap to evaluate spatial tracing. Experimental results show that RoboTracer surpasses baselines in spatial understanding, measuring, and referring, with an average success rate of 79.1%, and also achieves SOTA performance on TraceSpatial-Bench by a large margin, exceeding Gemini-2.5-Pro by 36% accuracy. Notably, RoboTracer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (UR5, G1 humanoid) in cluttered real-world scenes.
Authors: Kunhee Kim, NaHyeon Park, Kibeom Hong, Hyunjung Shim
Abstract: Textual Inversion (TI) is an efficient approach to text-to-image personalization but often fails on complex prompts. We trace these failures to embedding norm inflation: learned tokens drift to out-of-distribution magnitudes, degrading prompt conditioning in pre-norm Transformers. Empirically, we show semantics are primarily encoded by direction in CLIP token space, while inflated norms harm contextualization; theoretically, we analyze how large magnitudes attenuate positional information and hinder residual updates in pre-norm blocks. We propose Directional Textual Inversion (DTI), which fixes the embedding magnitude to an in-distribution scale and optimizes only direction on the unit hypersphere via Riemannian SGD. We cast direction learning as MAP with a von Mises-Fisher prior, yielding a constant-direction prior gradient that is simple and efficient to incorporate. Across personalization tasks, DTI improves text fidelity over TI and TI-variants while maintaining subject similarity. Crucially, DTI's hyperspherical parameterization enables smooth, semantically coherent interpolation between learned concepts (slerp), a capability that is absent in standard TI. Our findings suggest that direction-only optimization is a robust and scalable path for prompt-faithful personalization.
Authors: Jiachun Li, Jianan Feng, Jianjun Huang, Bin Liang
Abstract: Recently, object detection has proven vulnerable to adversarial patch attacks. The attackers holding a specially crafted patch can hide themselves from state-of-the-art detectors, e.g., YOLO, even in the physical world. This attack can bring serious security threats, such as escaping from surveillance cameras. How to effectively detect this kind of adversarial examples to catch potential attacks has become an important problem. In this paper, we propose two detection methods: the signature-based method and the signature-independent method. First, we identify two signatures of existing adversarial patches that can be utilized to precisely locate patches within adversarial examples. By employing the signatures, a fast signature-based method is developed to detect the adversarial objects. Second, we present a robust signature-independent method based on the \textit{content semantics consistency} of model outputs. Adversarial objects violate this consistency, appearing locally but disappearing globally, while benign ones remain consistently present. The experiments demonstrate that two proposed methods can effectively detect attacks both in the digital and physical world. These methods each offer distinct advantage. Specifically, the signature-based method is capable of real-time detection, while the signature-independent method can detect unknown adversarial patch attacks and makes defense-aware attacks almost impossible to perform.
Authors: Moein Sorkhei, Yue Liu, Hossein Azizpour, Edward Azavedo, Karin Dembrower, Dimitra Ntoula, Athanasios Zouzos, Fredrik Strand, Kevin Smith
Abstract: Interval and large invasive breast cancers, which are associated with worse prognosis than other cancers, are usually detected at a late stage due to false negative assessments of screening mammograms. The missed screening-time detection is commonly caused by the tumor being obscured by its surrounding breast tissues, a phenomenon called masking. To study and benchmark mammographic masking of cancer, in this work we introduce CSAW-M, the largest public mammographic dataset, collected from over 10,000 individuals and annotated with potential masking. In contrast to the previous approaches which measure breast image density as a proxy, our dataset directly provides annotations of masking potential assessments from five specialists. We also trained deep learning models on CSAW-M to estimate the masking level and showed that the estimated masking is significantly more predictive of screening participants diagnosed with interval and large invasive cancers -- without being explicitly trained for these tasks -- than its breast density counterparts.
Authors: Huu-Phu Do, Hao-Chien Hsueh, Tzu-Hao Chiang, Chi-Han Chen, Wen-Hsiao Peng, Ching-Chun Huang
Abstract: Photo exposure correction is widely investigated, but fewer studies focus on correcting under- and over-exposed images simultaneously. Three issues remain open to handle and correct both under- and over-exposed images in a unified way. First, a locally-adaptive exposure adjustment may be more flexible instead of learning a global mapping. Second, it is an ill-posed problem to determine the suitable exposure values locally. Third, photos with the same content but different exposures may not reach consistent adjustment results. To this end, we proposed a novel exposure correction network, ExReg, to address the challenges by formulating exposure correction as a multi-dimensional regression process. Given an input image, a compact multi-exposure generation network is introduced to generate images with different exposure conditions for multi-dimensional regression and exposure correction in the next stage. An auxiliary module is designed to predict the region-wise exposure values, guiding the proposed Encoder-Decoder ANP (Attentive Neural Processes) to regress the final corrected image. The experimental results show that ExReg can generate well-exposed results and outperform the SOTA method in PSNR for extensive exposure problems. Furthermore, the processing speed, with 0.05 seconds per image on an RTX 3090, is efficient. When tested on the same image under various exposure levels, ExReg also yields results that are visually consistent and physically accurate.
Authors: Xingjian Wang, Li Chai, Jiming Chen, Zhiguo Shi
Abstract: Multispectral pedestrian detection is essential to various tasks especially autonomous driving, for which both the accuracy and computational cost are of paramount importance. Most existing approaches treat RGB and infrared modalities equally. They typically adopt two symmetrical backbones for multimodal feature extraction, which ignore the substantial differences between modalities and bring great difficulty for the reduction of the computational cost as well as effective crossmodal fusion. In this work, we propose a novel and efficient framework named Wavelet-context Cooperative Network (WCCNet), which differentially extracts complementary features across spectra with low computational cost and further fuses these diverse features based on their spatially relevant cross-modal semantics. WCCNet explores an asymmetric but cooperative dual-stream backbone, in which WCCNet utilizes generic neural layers for texture-rich feature extraction from RGB modality, while proposing Mixture of Wavelet Experts (MoWE) to capture complementary frequency patterns of infrared modality. By assessing multispectral environmental context, MoWE generates routing scores to selectively activate specific learnable Adaptive DWT (ADWT) layers, alongside shared static DWT, which are both considerible lightwight and efficient to significantly reduce computational overhead and facilitate subsequent fusion. To further fuse these multispectral features with significant semantic differences, we elaborately design the crossmodal rearranging fusion module (CMRF), which aims to mitigate misalignment and merge semantically complementary features in spatially-related local regions to amplify the crossmodal reciprocal information. Results from comprehensive evaluations on KAIST and FLIR benchmarks indicate that WCCNet outperforms state-of-the-art methods with considerable computational efficiency and competitive accuracy.
Authors: Haorui Ji, Hongdong Li
Abstract: Diffusion models have demonstrated impressive capabilities in modeling complex data distributions and are increasingly applied in various generative tasks. In this work, we propose Pose Analysis by Diffusion Synthesis PADS, a unified generative modeling framework for 3D human pose analysis. PADS first learns a task-agnostic 3D pose prior via unconditional diffusion synthesis and then performs training-free adaptation to a wide range of pose analysis tasks, including 3D pose estimation, denoising, completion, etc., through a posterior sampling scheme. By formulating each task as an inverse problem with a known forward operator, PADS injects task-specific constraints during inference while keeping the pose prior fixed. This plug-and-play framework removes the need for task-specific supervision or retraining, offering flexibility and scalability across diverse conditions. Extensive experiments on different benchmarks showcase the superior performance against both learning-based and optimization-based baselines, demonstrating the effectiveness and generalization capability of our method.
Authors: Jean-Nicolas J\'er\'emie, Emmanuel Dauc\'e, Laurent U Perrinet
Abstract: From falcons spotting preys to humans recognizing faces, rapid visual abilities depend on a foveated retinal organization which delivers high-acuity central vision while preserving low-resolution periphery. This organization is conserved along early visual pathways but remains underexplored in machine learning. Here we examine how embedding a foveated retinotopic transformation as a preprocessing layer impacts convolutional neural networks (CNNs) for image classification. By applying a log-polar mapping to off-the-shelf models and retraining them, we retain comparable accuracy while improving robustness to scale and rotation. We show that this architecture becomes highly sensitive to fixation-point shifts, and that this sensitivity yields a proxy for defining saliency maps that effectively facilitates object localization. Our results show that foveated retinotopy encodes prior geometric knowledge, offering a solution to visual-search and enhancing both classification and localization. These findings connect biological vision principles with artificial networks, pointing to new, robust and efficient directions for computer-vision systems.
Authors: Shaoxu Li, Ye Pan
Abstract: Despite significant advancements in image generation using advanced generative frameworks, cross-image integration of content and style remains a key challenge. Current generative models, while powerful, frequently depend on vague textual prompts to define styles--creating difficulties in balancing content semantics and style preservation. We propose a novel framework that utilizes customized models to learn style representations. It enhances content preservation through cross-model feature and attention modulation, leveraging the inherent semantic consistency across models. Additionally, we introduce fixed feature and adaptive attention fusion to achieve the desired balance between content and style. We further develop spatial (mask-guided localized) and temporal (multi-style compositional) multi-model combinations, enabling flexible fusion of models and styles. Extensive experiments demonstrate that our method outperforms state-of-the-art approaches in balancing content preservation and stylistic coherence.
Authors: Jing Xu, Wentao Shi, Sheng Ren, Lijuan Zhang, Weikai Yang, Pan Gao, Jie Qin
Abstract: Effective monitoring of unusual transportation behaviors, such as wrong-way cycling (i.e., riding a bicycle or e-bike against designated traffic flow), is crucial for optimizing law enforcement deployment and traffic planning. However, accurately recording all wrong-way cycling events is both unnecessary and infeasible in resource-constrained environments, as it requires high-resolution cameras for evidence collection and event detection. To address this challenge, we propose WWC-Predictor, a novel method for efficiently estimating the wrong-way cycling ratio, defined as the proportion of wrong-way cycling events relative to the total number of cycling movements over a given time period. The core innovation of our method lies in accurately detecting wrong-way cycling events in sparsely sampled frames using a light-weight detector, then estimating the overall ratio using an autoregressive moving average model. To evaluate the effectiveness of our method, we construct a benchmark dataset consisting of 35 minutes of video sequences with minute-level annotations.Our method achieves an average error rate of a mere 1.475\% while consuming only 19.12\% GPU time required by conventional tracking methods, validating its effectiveness in estimating the wrong-way cycling ratio. Our source code is publicly available at: https://github.com/VICA-Lab-HKUST-GZ/WWC-Predictor.
Authors: Jiaben Chen, Xin Yan, Yihang Chen, Siyuan Cen, Zixin Wang, Qinwei Ma, Haoyu Zhen, Kaizhi Qian, Lie Lu, Chuang Gan
Abstract: In this work, we introduce a challenging task for simultaneously generating 3D holistic body motions and singing vocals directly from textual lyrics inputs, advancing beyond existing works that typically address these two modalities in isolation. To facilitate this, we first collect the RapVerse dataset, a large dataset containing synchronous rapping vocals, lyrics, and high-quality 3D holistic body meshes. With the RapVerse dataset, we investigate the extent to which scaling autoregressive multimodal transformers across language, audio, and motion can enhance the coherent and realistic generation of vocals and whole-body human motions. For modality unification, a vector-quantized variational autoencoder is employed to encode whole-body motion sequences into discrete motion tokens, while a vocal-to-unit model is leveraged to obtain quantized audio tokens preserving content, prosodic information and singer identity. By jointly performing transformer modeling on these three modalities in a unified way, our framework ensures a seamless and realistic blend of vocals and human motions. Extensive experiments demonstrate that our unified generation framework not only produces coherent and realistic singing vocals alongside human motions directly from textual inputs, but also rivals the performance of specialized single-modality generation systems, establishing new benchmarks for joint vocal-motion generation.
Authors: Wenfang Sun, Yingjun Du, Gaowen Liu, Yefeng Zheng, Cees G. M. Snoek
Abstract: We tackle the problem of quantifying the number of objects by a generative text-to-image model. Rather than retraining such a model for each new image domain of interest, which leads to high computational costs and limited scalability, we are the first to consider this problem from a domain-agnostic perspective. We propose QUOTA, an optimization framework for text-to-image models that enables effective object quantification across unseen domains without retraining. It leverages a dual-loop meta-learning strategy to optimize a domain-invariant prompt. Further, by integrating prompt learning with learnable counting and domain tokens, our method captures stylistic variations and maintains accuracy, even for object classes not encountered during training. For evaluation, we adopt a new benchmark specifically designed for object quantification in domain generalization, enabling rigorous assessment of object quantification accuracy and adaptability across unseen domains in text-to-image generation. Extensive experiments demonstrate that QUOTA outperforms conventional models in both object quantification accuracy and semantic consistency, setting a new benchmark for efficient and scalable text-to-image generation for any domain.
Authors: Dongwei Pan, Yang Li, Hongsheng Li, Kwan-Yee Lin
Abstract: We present TimeWalker, a novel framework that models realistic, full-scale 3D head avatars of a person on lifelong scale. Unlike current human head avatar pipelines that capture identity at the momentary level(e.g., instant photography or short videos), TimeWalker constructs a person's comprehensive identity from unstructured data collection over his/her various life stages, offering a paradigm to achieve full reconstruction and animation of that person at different moments of life. At the heart of TimeWalker's success is a novel neural parametric model that learns personalized representation with the disentanglement of shape, expression, and appearance across ages. Central to our methodology are the concepts of two aspects: (1) We track back to the principle of modeling a person's identity in an additive combination of average head representation in the canonical space, and moment-specific head attribute representations driven from a set of neural head basis. To learn the set of head basis that could represent the comprehensive head variations in a compact manner, we propose a Dynamic Neural Basis-Blending Module (Dynamo). It dynamically adjusts the number and blend weights of neural head bases, according to both shared and specific traits of the target person over ages. (2) Dynamic 2D Gaussian Splatting (DNA-2DGS), an extension of Gaussian splatting representation, to model head motion deformations like facial expressions without losing the realism of rendering and reconstruction. DNA-2DGS includes a set of controllable 2D oriented planar Gaussian disks that utilize the priors from parametric model, and move/rotate with the change of expression. Through extensive experimental evaluations, we show TimeWalker's ability to reconstruct and animate avatars across decoupled dimensions with realistic rendering effects, demonstrating a way to achieve personalized 'time traveling' in a breeze.
Authors: Nick Stracke, Stefan Andreas Baumann, Kolja Bauer, Frank Fundel, Bj\"orn Ommer
Abstract: Internal features from large-scale pre-trained diffusion models have recently been established as powerful semantic descriptors for a wide range of downstream tasks. Works that use these features generally need to add noise to images before passing them through the model to obtain the semantic features, as the models do not offer the most useful features when given images with little to no noise. We show that this noise has a critical impact on the usefulness of these features that cannot be remedied by ensembling with different random noises. We address this issue by introducing a lightweight, unsupervised fine-tuning method that enables diffusion backbones to provide high-quality, noise-free semantic features. We show that these features readily outperform previous diffusion features by a wide margin in a wide variety of extraction setups and downstream tasks, offering better performance than even ensemble-based methods at a fraction of the cost.
Authors: Biquard Maud, Marie Chabert, Florence Genin, Christophe Latry, Thomas Oberlin
Abstract: Satellite optical images, upon their on-ground receipt, offer a distorted view of the observed scene. Their restoration, including denoising, deblurring, and sometimes super-resolution, is required before their exploitation. Moreover, quantifying the uncertainties related to this restoration helps to reduce the risks of misinterpreting the image content. Deep learning methods are now state-of-the-art for satellite image restoration. Among them, direct inversion methods train a specific network for each sensor, and generally provide a point estimation of the restored image without the associated uncertainties. Alternatively, deep regularization (DR) methods learn a deep prior on target images before plugging it, as the regularization term, into a model-based optimization scheme. This allows for restoring images from several sensors with a single network and possibly for estimating associated uncertainties. In this paper, we introduce VBLE-xz, a DR method that solves the inverse problem in the latent space of a variational compressive autoencoder (CAE). We adapt the regularization strength by modulating the bitrate of the trained CAE with a training-free approach. Then, VBLE-xz estimates relevant uncertainties jointly in the latent and in the image spaces by sampling an explicit posterior estimated within variational inference. This enables fast posterior sampling, unlike state-of-the-art DR methods that use Markov chains or diffusion-based approaches. We conduct a comprehensive set of experiments on very high-resolution simulated and real Pl\'eiades images, asserting the performance, robustness and scalability of the proposed method. They demonstrate that VBLE-xz represents a compelling alternative to direct inversion methods when uncertainty quantification is required. The code associated to this paper is available in https://github.com/MaudBqrd/VBLExz.
Authors: Zijian Zhao, Zhijie Cai, Tingwei Chen, Xiaoyang Li, Hang Li, Qimei Chen, Guangxu Zhu
Abstract: Wireless sensing has recently found widespread applications in diverse environments, including homes, offices, and public spaces. By analyzing patterns in channel state information (CSI), it is possible to infer human actions for tasks such as person identification, gesture recognition, and fall detection. However, CSI is highly sensitive to environmental changes, where even minor alterations can significantly distort the CSI patterns. This sensitivity often leads to performance degradation or outright failure when applying wireless sensing models trained in one environment to another. To address this challenge, Domain Alignment (DAL) has been widely adopted for cross-domain classification tasks, as it focuses on aligning the global distributions of the source and target domains in feature space. Despite its popularity, DAL often neglects inter-category relationships, which can lead to misalignment between categories across domains, even when global alignment is achieved. To overcome these limitations, we propose K-Nearest Neighbors Maximum Mean Discrepancy (KNN-MMD), a novel few-shot method for cross-domain wireless sensing. Our approach begins by constructing a help set using KNN from the target domain, enabling local alignment between the source and target domains within each category using MMD. Additionally, we address a key instability issue commonly observed in cross-domain methods, where model performance fluctuates sharply between epochs. Further, most existing methods struggle to determine an optimal stopping point during training due to the absence of labeled data from the target domain. Our method resolves this by excluding the support set from the target domain during training and employing it as a validation set to determine the stopping criterion. The dataset and code are publicly available at https://github.com/RS2002/KNN-MMD .
Authors: Hanqing Jiang, Xiaojun Xiang, Han Sun, Hongjie Li, Liyang Zhou, Xiaoyu Zhang, Guofeng Zhang
Abstract: 3D Gaussian Splatting (3DGS) has recently attracted wide attentions in various areas such as 3D navigation, Virtual Reality (VR) and 3D simulation, due to its photorealistic and efficient rendering performance. High-quality reconstrution of 3DGS relies on sufficient splats and a reasonable distribution of these splats to fit real geometric surface and texture details, which turns out to be a challenging problem. We present GeoTexDensifier, a novel geometry-texture-aware densification strategy to reconstruct high-quality Gaussian splats which better comply with the geometric structure and texture richness of the scene. Specifically, our GeoTexDensifier framework carries out an auxiliary texture-aware densification method to produce a denser distribution of splats in fully textured areas, while keeping sparsity in low-texture regions to maintain the quality of Gaussian point cloud. Meanwhile, a geometry-aware splitting strategy takes depth and normal priors to guide the splitting sampling and filter out the noisy splats whose initial positions are far from the actual geometric surfaces they aim to fit, under a Validation of Depth Ratio Change checking. With the help of relative monocular depth prior, such geometry-aware validation can effectively reduce the influence of scattered Gaussians to the final rendering quality, especially in regions with weak textures or without sufficient training views. The texture-aware densification and geometry-aware splitting strategies are fully combined to obtain a set of high-quality Gaussian splats. We experiment our GeoTexDensifier framework on various datasets and compare our Novel View Synthesis results to other state-of-the-art 3DGS approaches, with detailed quantitative and qualitative evaluations to demonstrate the effectiveness of our method in producing more photorealistic 3DGS models.
Authors: Yikang Zhang, Chuang-Wei Liu, Jiahang Li, Yingbing Chen, Jie Cheng, Rui Fan
Abstract: Road inspection is crucial for maintaining road serviceability and ensuring traffic safety, as road defects gradually develop and compromise functionality. Traditional inspection methods, which rely on manual evaluations, are labor-intensive, costly, and time-consuming. While data-driven approaches are gaining traction, the scarcity and spatial sparsity of real-world road defects present significant challenges in acquiring high-quality datasets. Existing simulators designed to generate detailed synthetic driving scenes, however, lack models for road defects. Moreover, advanced driving tasks that involve interactions with road surfaces, such as planning and control in defective areas, remain underexplored. To address these limitations, we propose a multi-modal sensor platform integrated with an urban digital twin (UDT) system for intelligent road inspection. First, hierarchical road models are constructed from real-world driving data collected using vehicle-mounted sensors, resulting in highly detailed representations of road defect structures and surface elevations. Next, digital road twins are generated to create simulation environments for comprehensive analysis and evaluation of algorithm performance. These scenarios are then imported into a simulator to facilitate both data acquisition and physical simulation. Experimental results demonstrate that driving tasks, including perception and decision-making, benefit significantly from the high-fidelity road defect scenes generated by our system.
Authors: Haorui Ji, Taojun Lin, Hongdong Li
Abstract: Diffusion models demonstrate remarkable capabilities in capturing complex data distributions and have achieved compelling results in many generative tasks. While they have recently been extended to dense prediction tasks such as depth estimation and surface normal prediction, their full potential in this area remains underexplored. As target signal maps and input images are pixel-wise aligned, the conventional noise-to-data generation paradigm is inefficient, and input images can serve as a more informative prior compared to pure noise. Diffusion bridge models, which support data-to-data generation between two general data distributions, offer a promising alternative, but they typically fail to exploit the rich visual priors embedded in large pretrained foundation models. To address these limitations, we integrate diffusion bridge formulation with structured visual priors and introduce DPBridge, the first latent diffusion bridge framework for dense prediction tasks. To resolve the incompatibility between diffusion bridge models and pretrained diffusion backbones, we propose (1) a tractable reverse transition kernel for the diffusion bridge process, enabling maximum likelihood training scheme; (2) finetuning strategies including distribution-aligned normalization and image consistency loss. Experiments across extensive benchmarks validate that our method consistently achieves superior performance, demonstrating its effectiveness and generalization capability under different scenarios.
Authors: Souvik Maji, Rhythm Baghel, Pratik Mazumder
Abstract: Few-shot learning has been extensively explored to address problems where the amount of labeled samples is very limited for some classes. In the semi-supervised few-shot learning setting, substantial quantities of unlabeled samples are available. Such unlabeled samples are generally cheaper to obtain and can be used to improve the few-shot learning performance of the model. Some of the recent methods for this setting rely on clustering to generate pseudo-labels for the unlabeled samples. Since the effectiveness of clustering heavily influences the labeling of the unlabeled samples, it can significantly affect the few-shot learning performance. In this paper, we focus on improving the representation learned by the model in order to improve the clustering and, consequently, the model performance. We propose an approach for semi-supervised few-shot learning that performs a class-variance optimized clustering coupled with a cluster separation tuner in order to improve the effectiveness of clustering the labeled and unlabeled samples in this setting. It also optimizes the clustering-based pseudo-labeling process using a restricted pseudo-labeling approach and performs semantic information injection in order to improve the semi-supervised few-shot learning performance of the model. We experimentally demonstrate that our proposed approach significantly outperforms recent state-of-the-art methods on the benchmark datasets. To further establish its robustness, we conduct extensive experiments under challenging conditions, showing that the model generalizes well to domain shifts and achieves new state-of-the-art performance in open-set settings with distractor classes, highlighting its effectiveness for real-world applications.
Authors: Tim Fl\"uckiger, Jonas Hein, Valery Fischer, Philipp F\"urnstahl, Lilian Calvet
Abstract: The purpose of this study is to develop an automated and accurate external camera calibration method for multi-camera systems used in 3D surgical scene reconstruction (3D-SSR), eliminating the need for operator intervention or specialized expertise. The method specifically addresses the problem of limited overlapping fields of view caused by significant variations in optical zoom levels and camera locations. We contribute a novel, fast, and fully automatic calibration method based on the projection of multi-scale markers (MSMs) using a ceiling-mounted projector. MSMs consist of 2D patterns projected at varying scales, ensuring accurate extraction of well distributed point correspondences across significantly different viewpoints and zoom levels. Validation is performed using both synthetic and real data captured in a mock-up OR, with comparisons to traditional manual marker-based methods as well as markerless calibration methods. The method achieves accuracy comparable to manual, operator-dependent calibration methods while exhibiting higher robustness under conditions of significant differences in zoom levels. Additionally, we show that state-of-the-art Structure-from-Motion (SfM) pipelines are ineffective in 3D-SSR settings, even when additional texture is projected onto the OR floor. The use of a ceiling-mounted entry-level projector proves to be an effective alternative to operator-dependent, traditional marker-based methods, paving the way for fully automated 3D-SSR.
Authors: Wanke Xia, Ruoxin Peng, Haoqi Chu, Xinlei Zhu, Zhiyu Yang, Yiting Zhao, Lili Yang
Abstract: Rice is one of the most widely cultivated crops globally and has been developed into numerous varieties. The quality of rice during cultivation is primarily determined by its cultivar and characteristics. Traditionally, rice classification and quality assessment rely on manual visual inspection, a process that is both time-consuming and prone to errors. However, with advancements in machine vision technology, automating rice classification and quality evaluation based on its cultivar and characteristics has become increasingly feasible, enhancing both accuracy and efficiency. This study proposes a real-time evaluation mechanism for comprehensive rice grain assessment, integrating a one-stage object detection approach, a deep convolutional neural network, and traditional machine learning techniques. The proposed framework enables rice variety identification, grain completeness grading, and grain chalkiness evaluation. The rice grain dataset used in this study comprises approximately 20,000 images from six widely cultivated rice varieties in China. Experimental results demonstrate that the proposed mechanism achieves a mean average precision (mAP) of 99.14% in the object detection task and an accuracy of 97.89% in the classification task. Furthermore, the framework attains an average accuracy of 97.56% in grain completeness grading within the same rice variety, contributing to an effective quality evaluation system.
Authors: Wanke Xia, Ruoxin Peng, Haoqi Chu, Xinlei Zhu, Zhiyu Yang, Lili Yang, Bo Lv, Xunwen Xiang
Abstract: Rice is a staple food for a significant portion of the world's population, providing essential nutrients and serving as a versatile in-gredient in a wide range of culinary traditions. Recently, the use of deep learning has enabled automated classification of rice, im-proving accuracy and efficiency. However, classical models based on first-stage training may face difficulties in distinguishing between rice varieties with similar external characteristics, thus leading to misclassifications. Considering the transparency and feasibility of model, we selected and gradually improved pure fully connected neural network to achieve classification of rice grain. The dataset we used contains both global and domestic rice images obtained from websites and laboratories respectively. First, the training mode was changed from one-stage training to two-stage training, which significantly contributes to distinguishing two similar types of rice. Secondly, the preprocessing method was changed from random tilting to horizontal or vertical position cor-rection. After those two enhancements, the accuracy of our model increased notably from 97% to 99%. In summary, two subtle methods proposed in this study can remarkably enhance the classification ability of deep learning models in terms of the classification of rice grain.
Authors: Jun Yang, Wenjie Xue, Sahar Ghavidel, Steven L. Waslander
Abstract: Estimating the 6D pose of textureless objects from RGB images is an important problem in robotics. Due to appearance ambiguities, rotational symmetries, and severe occlusions, single-view based 6D pose estimators are still unable to handle a wide range of objects, motivating research towards multi-view pose estimation and next-best-view prediction that addresses these limitations. In this work, we propose a comprehensive active perception framework for estimating the 6D poses of textureless objects using only RGB images. Our approach is built upon a key idea: decoupling the 6D pose estimation into a two-step sequential process can greatly improve both accuracy and efficiency. First, we estimate the 3D translation of each object, resolving scale and depth ambiguities inherent to RGB images. These estimates are then used to simplify the subsequent task of determining the 3D orientation, which we achieve through canonical scale template matching. Building on this formulation, we then introduce an active perception strategy that predicts the next best camera viewpoint to capture an RGB image, effectively reducing object pose uncertainty and enhancing pose accuracy. We evaluate our method on the public ROBI and TOD datasets, as well as on our reconstructed transparent object dataset, T-ROBI. Under the same camera viewpoints, our multi-view pose estimation significantly outperforms state-of-the-art approaches. Furthermore, by leveraging our next-best-view strategy, our approach achieves high pose accuracy with fewer viewpoints than heuristic-based policies across all evaluated datasets. The accompanying video and T-ROBI dataset will be released on our project page: https://trailab.github.io/ActiveODPE.
Authors: Jianxiong Gao, Yichang Liu, Baofeng Yang, Jianfeng Feng, Yanwei Fu
Abstract: Most research decoding brain signals into images, often using them as priors for generative models, has focused only on visual content. This overlooks the brain's natural ability to integrate auditory and visual information, for instance, sound strongly influences how we perceive visual scenes. To investigate this, we propose a new task of reconstructing continuous video stimuli from multimodal brain signals recorded during audiovisual stimulation. To enable this, we introduce CineBrain, the first large-scale dataset that synchronizes fMRI and EEG during audiovisual viewing, featuring six hours of \textit{The Big Bang Theory} episodes for cross-modal alignment. We also conduct the first systematic exploration of combining fMRI and EEG for video reconstruction and present CineSync, a framework for reconstructing dynamic video using a Multi-Modal Fusion Encoder and a Neural Latent Decoder. CineSync achieves state-of-the-art performance in dynamic reconstruction, leveraging the complementary strengths of fMRI and EEG to improve visual fidelity. Our analysis shows that auditory cortical activations enhance decoding accuracy, highlighting the role of auditory input in visual perception. Project Page: https://jianxgao.github.io/CineBrain.
Authors: Chenyu Zhang, Yihao Luo, Lei Zhu, Martyn G Boutelle, Choon Hwai Yap, Guang Yang
Abstract: Accurate reconstruction of multi-chamber cardiac anatomy from medical images is a cornerstone for patient-specific modeling, physiological simulation, and interventional planning. However, current reconstruction pipelines fundamentally rely on surface-wise geometric supervision and model each chamber in isolation, resulting in anatomically implausible inter-chamber violations despite apparently favorable overlap or distance metrics. In this work, we propose a relational anatomical supervision framework for multi-chamber cardiac mesh reconstruction by introducing a Mesh Interrelation Enhancement (MIE) loss. The proposed formulation explicitly encodes spatial relationships between cardiac structures into a differentiable occupancy-based objective, thereby transforming qualitative anatomical rules into quantitative geometric supervision. We further establish violation-aware evaluation metrics to directly quantify inter-chamber structural correctness, revealing systematic limitations of commonly used geometric measures such as Dice and Chamfer distance. Extensive experiments on multi-center CT data, densely sampled MR data, and two independent external cohorts, including a highly heterogeneous congenital heart disease population, demonstrate that the proposed method consistently suppresses clinically critical boundary violations by up to 83\%, while maintaining competitive volumetric accuracy and achieving superior surface fidelity. Notably, the proposed relational supervision generalizes robustly across imaging modalities, centers, and pathological conditions, even under severe anatomical deformation. These results demonstrate that distance-based supervision alone is insufficient to guarantee anatomically faithful reconstruction, and that explicit enforcement of multi-structure anatomical relations provides a principled and robust pathway toward reliable patient-specific cardiac modeling.
Authors: Parsa Hosseini, Sumit Nawathe, Mazda Moayeri, Sriram Balasubramanian, Soheil Feizi
Abstract: Unimodal vision models are known to rely on spurious correlations, but it remains unclear to what extent Multimodal Large Language Models (MLLMs) exhibit similar biases despite language supervision. In this paper, we investigate spurious bias in MLLMs and introduce SpurLens, a pipeline that leverages GPT-4 and open-set object detectors to automatically identify spurious visual cues without human supervision. Our findings reveal that spurious correlations cause two major failure modes in MLLMs: (1) over-reliance on spurious cues for object recognition, where removing these cues reduces accuracy, and (2) object hallucination, where spurious cues amplify the hallucination by over 10x. We validate our findings in various MLLMs and datasets. Beyond diagnosing these failures, we explore potential mitigation strategies, such as prompt ensembling and reasoning-based prompting, and conduct ablation studies to examine the root causes of spurious bias in MLLMs. By exposing the persistence of spurious correlations, our study calls for more rigorous evaluation methods and mitigation strategies to enhance the reliability of MLLMs.
Authors: Wenjie Li, Heng Guo, Yuefeng Hou, Zhanyu Ma
Abstract: Image super-resolution (SR) aims to recover low-resolution images to high-resolution images, where improving SR efficiency is a high-profile challenge. However, commonly used units in SR, like convolutions and window-based Transformers, have limited receptive fields, making it challenging to apply them to improve SR under extremely limited computational cost. To address this issue, inspired by modeling convolution theorem through token mix, we propose a Fourier token-based plugin called FourierSR to improve SR uniformly, which avoids the instability or inefficiency of existing token mix technologies when applied as plug-ins. Furthermore, compared to convolutions and windows-based Transformers, our FourierSR only utilizes Fourier transform and multiplication operations, greatly reducing complexity while having global receptive fields. Experimental results show that our FourierSR as a plug-and-play unit brings an average PSNR gain of 0.34dB for existing efficient SR methods on Manga109 test set at the scale of x4, while the average increase in the number of Params and FLOPs is only 0.6% and 1.5% of original sizes. We will release our codes upon acceptance.
Authors: Han Liu, Riqiang Gao, Eileen Krieg, Sasa Grbic
Abstract: Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive forms of pancreatic cancer and is often diagnosed at an advanced stage due to subtle early imaging signs. To enable earlier detection and improve clinical decision-making, we propose a coarse-to-fine AI-assisted framework named PanDx for identifying PDAC on contrast-enhanced CT scans. Our approach integrates two novel techniques: (1) distribution-aware stratified ensembling to improve generalization across lesion variations, and (2) peak-scaled lesion candidate extraction to enhance lesion localization precision. PanDx is developed and evaluated as part of the PANORAMA challenge, where it ranked 1st place on the official test set with an AUROC of 0.9263 and an AP of 0.7243. Furthermore, we analyzed failure cases with a radiologist to identify the limitations of AI models on this task and discussed potential future directions for model improvement. Our code and models are publicly available at https://github.com/han-liu/PanDx.
Authors: Leqi Shen, Guoqiang Gong, Tao He, Yifeng Zhang, Pengzhang Liu, Sicheng Zhao, Guiguang Ding
Abstract: Video Large Language Models have demonstrated strong video understanding capabilities, yet their practical deployment is hindered by substantial inference costs caused by redundant video tokens. Existing pruning techniques fail to effectively exploit the spatiotemporal redundancy present in video data. To bridge this gap, we perform a systematic analysis of video redundancy from two perspectives: temporal context and visual context. Leveraging these insights, we propose Dynamic Density Pruning for Fast Video LLMs termed FastVID. Specifically, FastVID dynamically partitions videos into temporally ordered segments to preserve temporal structure and applies a density-based token pruning strategy to maintain essential spatial and temporal information. Our method significantly reduces computational overhead while maintaining temporal and visual integrity. Extensive evaluations show that FastVID achieves state-of-the-art performance across various short- and long-video benchmarks on leading Video LLMs, including LLaVA-OneVision, LLaVA-Video, Qwen2-VL, and Qwen2.5-VL. Notably, on LLaVA-OneVision-7B, FastVID effectively prunes $\textbf{90.3%}$ of video tokens, reduces FLOPs to $\textbf{8.3%}$, and accelerates the LLM prefill stage by $\textbf{7.1}\times$, while maintaining $\textbf{98.0%}$ of the original accuracy. The code is available at https://github.com/LunarShen/FastVID.
Authors: Susung Hong, Ira Kemelmacher-Shlizerman, Brian Curless, Steven M. Seitz
Abstract: We introduce MusicInfuser, an approach that aligns pre-trained text-to-video diffusion models to generate high-quality dance videos synchronized with specified music tracks. Rather than training a multimodal audio-video or audio-motion model from scratch, our method demonstrates how existing video diffusion models can be efficiently adapted to align with musical inputs. We propose a novel layer-wise adaptability criterion based on a guidance-inspired constructive influence function to select adaptable layers, significantly reducing training costs while preserving rich prior knowledge, even with limited, specialized datasets. Experiments show that MusicInfuser effectively bridges the gap between music and video, generating novel and diverse dance movements that respond dynamically to music. Furthermore, our framework generalizes well to unseen music tracks, longer video sequences, and unconventional subjects, outperforming baseline models in consistency and synchronization. All of this is achieved without requiring motion data, with training completed on a single GPU within a day.
Authors: Jae-Young Yim, Dongwook Kim, Jae-Young Sim
Abstract: Large-scale datasets are usually required to train deep neural networks, but it increases the computational complexity hindering the practical applications. Recently, dataset distillation for images and texts has been attracting a lot of attention, that reduces the original dataset to a synthetic dataset to alleviate the computational burden of training while preserving essential task-relevant information. However, the dataset distillation for 3D point clouds remains largely unexplored, as the point clouds exhibit fundamentally different characteristics from that of images, making the dataset distillation more challenging. In this paper, we propose a distribution matching-based distillation framework for 3D point clouds that jointly optimizes the geometric structures as well as the orientations of the synthetic 3D objects. To address the semantic misalignment caused by unordered indexing of points, we introduce a Semantically Aligned Distribution Matching loss computed on the sorted features in each channel. Moreover, to address the rotation variation, we jointly learn the optimal rotation angles while updating the synthetic dataset to better align with the original feature distribution. Extensive experiments on widely used benchmark datasets demonstrate that the proposed method consistently outperforms existing dataset distillation methods, achieving superior accuracy and strong cross-architecture generalization.
Authors: Zhong-Yu Li, Ruoyi Du, Juncheng Yan, Le Zhuo, Qilong Wu, Zhen Li, Peng Gao, Zhanyu Ma, Ming-Ming Cheng
Abstract: Recent progress in diffusion models significantly advances various image generation tasks. However, the current mainstream approach remains focused on building task-specific models, which have limited efficiency when supporting a wide range of different needs. While universal models attempt to address this limitation, they face critical challenges, including generalizable task instruction, appropriate task distributions, and unified architectural design. To tackle these challenges, we propose VisualCloze, a universal image generation framework, which supports a wide range of in-domain tasks, generalization to unseen ones, unseen unification of multiple tasks, and reverse generation. Unlike existing methods that rely on language-based task instruction, leading to task ambiguity and weak generalization, we integrate visual in-context learning, allowing models to identify tasks from visual demonstrations. Meanwhile, the inherent sparsity of visual task distributions hampers the learning of transferable knowledge across tasks. To this end, we introduce Graph200K, a graph-structured dataset that establishes various interrelated tasks, enhancing task density and transferable knowledge. Furthermore, we uncover that our unified image generation formulation shared a consistent objective with image infilling, enabling us to leverage the strong generative priors of pre-trained infilling models without modifying the architectures.
Authors: Lin Li, Wei Chen, Jiahui Li, Kwang-Ting Cheng, Long Chen
Abstract: Recent advances in multi-modal large language models (MLLMs) have significantly improved object-level grounding and region captioning. However, they remain limited in visual relation understanding, struggling even with binary relation detection, let alone \textit{N}-ary relations involving multiple semantic roles. The core reason is the lack of modeling for \textit{structural semantic dependencies} among multi-entities, leading to unreliable outputs, hallucinations, and over-reliance on language priors (\eg, defaulting to ``person drinks a milk'' if a person is merely holding it). To this end, we propose Relation-R1, the \textit{first unified} relation comprehension framework that explicitly integrates cognitive chain-of-thought (CoT)-guided supervised fine-tuning (SFT) and group relative policy optimization (GRPO) within a reinforcement learning (RL) paradigm. Specifically, we first establish foundational reasoning capabilities via SFT, enforcing structured outputs with thinking processes. Then, GRPO is utilized to refine these outputs via multi-rewards optimization, prioritizing visual-semantic grounding over language-induced biases, thereby improving generalization capability. Furthermore, we investigate the impact of various CoT strategies within this framework, demonstrating that a specific-to-general progressive approach in CoT guidance further improves generalization, especially in capturing synonymous \textit{N}-ary relations. Extensive experiments on widely-used PSG and SWiG datasets demonstrate that Relation-R1 achieves state-of-the-art performance in both binary and \textit{N}-ary relation understanding.
Authors: Xiao Zhang, Lu Zou, Tao Lu, Yuan Yao, Zhangjin Huang, Guoping Wang
Abstract: Category-level object pose estimation aims to predict the 6D pose and size of previously unseen instances from predefined categories, requiring strong generalization across diverse object instances. Although many previous methods attempt to mitigate intra-class variations, they often struggle with instances exhibiting complex geometries or significant deviations from canonical shapes. To address this issue, we propose INKL-Pose, a novel category-level object pose estimation framework that enables INstance-adaptive Keypoint Learning with local-to-global geometric aggregation. Specifically, our method first predicts semantically consistent and geometrically informative keypoints using an Instance-Adaptive Keypoint Detector, then refines them: (1) a Local Keypoint Feature Aggregator capturing fine-grained geometries, and (2) a Global Keypoint Feature Aggregator using bidirectional Mamba for structural consistency. To enable bidirectional modeling in Mamba, we introduce a simple yet effective Feature Sequence Flipping strategy that preserves spatial coherence while constructing backward feature sequence. Additionally, we design a surface loss and a separation loss to encourage uniform coverage and spatial diversity in keypoint distribution. The resulting keypoints are mapped to a canonical space for 6D pose and size regression. Extensive experiments on CAMERA25, REAL275, and HouseCat6D show that INKL-Pose achieves state-of-the-art performance with 16.7M parameters and runs at 36 FPS on an NVIDIA RTX 4090D GPU.
Authors: Keyang Ye, Tianjia Shao, Kun Zhou
Abstract: We introduce Gaussian-enhanced Surfels (GESs), a bi-scale representation for radiance field rendering, wherein a set of 2D opaque surfels with view-dependent colors represent the coarse-scale geometry and appearance of scenes, and a few 3D Gaussians surrounding the surfels supplement fine-scale appearance details. The rendering with GESs consists of two passes -- surfels are first rasterized through a standard graphics pipeline to produce depth and color maps, and then Gaussians are splatted with depth testing and color accumulation on each pixel order independently. The optimization of GESs from multi-view images is performed through an elaborate coarse-to-fine procedure, faithfully capturing rich scene appearance. The entirely sorting-free rendering of GESs not only achieves very fast rates, but also produces view-consistent images, successfully avoiding popping artifacts under view changes. The basic GES representation can be easily extended to achieve anti-aliasing in rendering (Mip-GES), boosted rendering speeds (Speedy-GES) and compact storage (Compact-GES), and reconstruct better scene geometries by replacing 3D Gaussians with 2D Gaussians (2D-GES). Experimental results show that GESs advance the state-of-the-arts as a compelling representation for ultra-fast high-fidelity radiance field rendering.
Authors: Mingrui Chen, Haogeng Liu, Hao Liang, Huaibo Huang, Wentao Zhang, Ran He
Abstract: In this work, we investigate how explicitly modeling problem's difficulty prior information shapes the effectiveness of reinforcement learning based fine-tuning for multimodal reasoning. Our exploration mainly comprises of following three perspective: First, through offline data curation, we analyze the U-shaped difficulty distribution of two given datasets using the base model by multi-round sampling, and then filter out prompts that are either too simple or extremely difficult to provide meaningful gradients and perform subsequent two-stage training. Second, we implement an online advantage differentiation, computing group-wise empirical accuracy as a difficulty proxy to adaptively reweight advantages estimation, providing stronger learning signals for more challenging problems. Finally, we introduce difficulty hints as explicit prompts for more complex samples in the second training stage, encouraging the model to calibrate its reasoning depth and perform reflective validation checks. Our comprehensive approach demonstrates significant performances across various multi-modal mathematical reasoning benchmarks with only 2K+0.6K two-stage training data.
Authors: Anthony Gosselin, Ge Ya Luo, Luis Lara, Florian Golemo, Derek Nowrouzezahrai, Liam Paull, Alexia Jolicoeur-Martineau, Christopher Pal
Abstract: Video diffusion techniques have advanced significantly in recent years; however, they struggle to generate realistic imagery of car crashes due to the scarcity of accident events in most driving datasets. Improving traffic safety requires realistic and controllable accident simulations. To tackle the problem, we propose Ctrl-Crash, a controllable car crash video generation model that conditions on signals such as bounding boxes, crash types, and an initial image frame. Our approach enables counterfactual scenario generation where minor variations in input can lead to dramatically different crash outcomes. To support fine-grained control at inference time, we leverage classifier-free guidance with independently tunable scales for each conditioning signal. Ctrl-Crash achieves state-of-the-art performance across quantitative video quality metrics (e.g., FVD and JEDi) and qualitative measurements based on a human-evaluation of physical realism and video quality compared to prior diffusion-based methods.
Authors: Melika Ayoughi, Samira Abnar, Chen Huang, Chris Sandino, Sayeri Lala, Eeshan Gunesh Dhekane, Dan Busbridge, Shuangfei Zhai, Vimal Thilak, Josh Susskind, Pascal Mettes, Paul Groth, Hanlin Goh
Abstract: The composition of objects and their parts, along with object-object positional relationships, provides a rich source of information for representation learning. Hence, spatial-aware pretext tasks have been actively explored in self-supervised learning. Existing works commonly start from a grid structure, where the goal of the pretext task involves predicting the absolute position index of patches within a fixed grid. However, grid-based approaches fall short of capturing the fluid and continuous nature of real-world object compositions. We introduce PART, a self-supervised learning approach that leverages continuous relative transformations between off-grid patches to overcome these limitations. By modeling how parts relate to each other in a continuous space, PART learns the relative composition of images-an off-grid structural relative positioning that is less tied to absolute appearance and can remain coherent under variations such as partial visibility or stylistic changes. In tasks requiring precise spatial understanding such as object detection and time series prediction, PART outperforms grid-based methods like MAE and DropPos, while maintaining competitive performance on global classification tasks. By breaking free from grid constraints, PART opens up a new trajectory for universal self-supervised pretraining across diverse datatypes-from images to EEG signals-with potential in medical imaging, video, and audio.
Authors: Gustavo Perez, Stella X. Yu
Abstract: Classical image filters, such as those for averaging or differencing, are carefully normalized to ensure consistency, interpretability, and to avoid artifacts like intensity shifts, halos, or ringing. In contrast, convolutional filters learned end-to-end in deep networks lack such constraints. Although they may resemble wavelets and blob/edge detectors, they are not normalized in the same or any way. Consequently, when images undergo atmospheric transfer, their responses become distorted, leading to incorrect outcomes. We address this limitation by proposing filter normalization, followed by learnable scaling and shifting, akin to batch normalization. This simple yet effective modification ensures that the filters are atmosphere-equivariant, enabling co-domain symmetry. By integrating classical filtering principles into deep learning (applicable to both convolutional neural networks and convolution-dependent vision transformers), our method achieves significant improvements on artificial and natural intensity variation benchmarks. Our ResNet34 could even outperform CLIP by a large margin. Our analysis reveals that unnormalized filters degrade performance, whereas filter normalization regularizes learning, promotes diversity, and improves robustness and generalization.
Authors: Marwane Kzadri, Franco Alberto Cardillo, Nan\'ee Chahinian, Carole Delenne, Renaud Hostache, Jamal Riffi
Abstract: Segmenting Synthetic Aperture Radar (SAR) images is crucial for many remote sensing applications, particularly water body detection. However, deep learning-based segmentation models often face challenges related to convergence speed and stability, mainly due to the complex statistical distribution of this type of data. In this study, we evaluate the impact of mode normalization on two widely used semantic segmentation models, U-Net and SegNet. Specifically, we integrate mode normalization, to reduce convergence time while maintaining the performance of the baseline models. Experimental results demonstrate that mode normalization significantly accelerates convergence. Furthermore, cross-validation results indicate that normalized models exhibit increased stability in different zones. These findings highlight the effectiveness of normalization in improving computational efficiency and generalization in SAR image segmentation.
Authors: Tianxiang Hao, Lixian Zhang, Yingjia Zhang, Mengxuan Chen, Jinxiao Zhang, Runmin Dong, Haohuan Fu
Abstract: Historical satellite imagery archive, such as Keyhole satellite data, offers rare insights into understanding early urban development and long-term transformation. However, severe quality degradation ($\textit{e.g.}$, distortion, misalignment, and spectral scarcity) and the absence of annotations have long hindered its analysis. To bridge this gap and enhance understanding of urban development, we introduce $\textbf{WakeupUrbanBench}$, an annotated segmentation dataset based on historical satellite imagery with the earliest observation time among all existing remote sensing (RS) datasets, along with a framework for unsupervised segmentation tasks, $\textbf{WakeupUSM}$. First, WakeupUrbanBench serves as a pioneer, expertly annotated dataset built on mid-$20^{\text{th}}$ century RS imagery, involving four key urban classes and spanning 4 cities across 2 continents with nearly 1000 km$^2$ area of diverse urban morphologies, and additionally introducing one present-day city. Second, WakeupUSM is a novel unsupervised semantic segmentation framework for historical RS imagery. It employs a confidence-aware alignment mechanism and focal-confidence loss based on a self-supervised learning architecture, which generates robust pseudo-labels and adaptively prioritizes prediction difficulty and label reliability to improve unsupervised segmentation on noisy historical data without manual supervision. Comprehensive experiments demonstrate WakeupUSM significantly outperforms existing unsupervised segmentation methods $\textbf{both WakeupUrbanBench and public dataset}$, promising to pave the way for quantitative studies of long-term urban change using modern computer vision. Our benchmark and codes will be released at https://github.com/Tianxiang-Hao/WakeupUrban.
Authors: Numair Nadeem, Saeed Anwar, Muhammad Hamza Asad, Abdul Bais
Abstract: Vision Language Models (VLMs) provide rich semantic priors but are underexplored in Semi supervised Semantic Segmentation. Recent attempts to integrate VLMs to inject high level semantics overlook the semantic misalignment between visual and textual representations that arises from using domain invariant text embeddings without adapting them to dataset and image specific contexts. This lack of domain awareness, coupled with limited annotations, weakens the model semantic understanding by preventing effective vision language alignment. As a result, the model struggles with contextual reasoning, shows weak intra class discrimination, and confuses similar classes. To address these challenges, we propose Hierarchical Vision Language transFormer (HVLFormer), which achieves domain aware and domain robust alignment between visual and textual representations within a mask transformer architecture. Firstly, we transform text embeddings from pretrained VLMs into textual object queries, enabling the generation of multi scale, dataset aware queries that capture class semantics from coarse to fine granularity and enhance contextual reasoning. Next, we refine these queries by injecting image specific visual context to align textual semantics with local scene structures and enhance class discrimination. Finally, to achieve domain robustness, we introduce cross view and modal consistency regularization, which enforces prediction consistency within mask-transformer architecture across augmented views. Moreover, it ensures stable vision language alignment during decoding. With less than 1% training data, HVLFormer outperforms state of the art methods on Pascal VOC, COCO, ADE20K, and Cityscapes. Our code and results will be available on GitHub.
Authors: Rim Assouel, Declan Campbell, Yoshua Bengio, Taylor Webb
Abstract: To accurately process a visual scene, observers must bind features together to represent individual objects. This capacity is necessary, for instance, to distinguish an image containing a red square and a blue circle from an image containing a blue square and a red circle. Recent work has found that language models solve this 'binding problem' via a set of symbol-like, content-independent indices, but it is unclear whether similar mechanisms are employed by Vision Language Models (VLMs). This question is especially relevant, given the persistent failures of VLMs on tasks that require binding. Here, we identify a previously unknown set of emergent symbolic mechanisms that support binding specifically in VLMs, via a content-independent, spatial indexing scheme. Moreover, we find that binding errors, when they occur, can be traced directly to failures in these mechanisms. Taken together, these results shed light on the mechanisms that support symbol-like processing in VLMs, and suggest possible avenues for reducing the number of binding failures exhibited by these models.
Authors: Chaehyeon Song, Dongjae Lee, Jongwoo Lim, Ayoung Kim
Abstract: Camera calibration using planar targets has been widely favored, and two types of control points have been mainly considered as measurements: the corners of the checkerboard and the centroid of circles. Since a centroid is derived from numerous pixels, the circular pattern provides more precise measurements than the checkerboard. However, the existing projection model of circle centroids is biased under lens distortion, resulting in low performance. To surmount this limitation, we propose an unbiased projection model of the circular pattern and demonstrate its superior accuracy compared to the checkerboard. Complementing this, we introduce uncertainty into circular patterns to enhance calibration robustness and completeness. Defining centroid uncertainty improves the performance of calibration components, including pattern detection, optimization, and evaluation metrics. We also provide guidelines for performing good camera calibration based on the evaluation metric. The core concept of this approach is to model the boundary points of a two-dimensional shape as a Markov random field, considering its connectivity. The shape distribution is propagated to the centroid uncertainty through an appropriate shape representation based on the Green theorem. Consequently, the resulting framework achieves marked gains in calibration accuracy and robustness. The complete source code and demonstration video are available at https://github.com/chaehyeonsong/discocal.
Authors: Chade Li, Pengju Zhang, Yihong Wu
Abstract: The rapid advancement of vision-language models (VLMs) in 3D domains has accelerated research in text-query-guided point cloud processing, though existing methods underperform in point-level segmentation due to inadequate 3D-text alignment that limits local feature-text context linking. To address this limitation, we propose MR-COSMO, a Visual-Text Memory Recall and Direct CrOSs-MOdal Alignment Method for Query-Driven 3D Segmentation, establishing explicit alignment between 3D point clouds and text/2D image data through a dedicated direct cross-modal alignment module while implementing a visual-text memory module with specialized feature banks. This direct alignment mechanism enables precise fusion of geometric and semantic features, while the memory module employs specialized banks storing text features, visual features, and their correspondence mappings to dynamically enhance scene-specific representations via attention-based knowledge recall. Comprehensive experiments across 3D instruction, reference, and semantic segmentation benchmarks confirm state-of-the-art performance.
Authors: Yansong Peng, Kai Zhu, Yu Liu, Pingyu Wu, Hebei Li, Xiaoyan Sun, Feng Wu
Abstract: Continuous-time Consistency Models (CMs) promise efficient few-step generation but face significant challenges with training instability. We argue this instability stems from a fundamental conflict: Training the network exclusively on a shortcut objective leads to the catastrophic forgetting of the instantaneous velocity field that defines the flow. Our solution is to explicitly anchor the model in the underlying flow, ensuring high trajectory fidelity during training. We introduce the Flow-Anchored Consistency Model (FACM), where a Flow Matching (FM) task serves as a dynamic anchor for the primary CM shortcut objective. Key to this Flow-Anchoring approach is a novel expanded time interval strategy that unifies optimization for a single model while decoupling the two tasks to ensure stable, architecturally-agnostic training. By distilling a pre-trained LightningDiT model, our method achieves a state-of-the-art FID of 1.32 with two steps (NFE=2) and 1.70 with just one step (NFE=1) on ImageNet 256x256. To address the challenge of scalability, we develop a memory-efficient Chain-JVP that resolves key incompatibilities with FSDP. This method allows us to scale FACM training on a 14B parameter model (Wan 2.2), accelerating its Text-to-Image inference from 2x40 to 2-8 steps. Our code and pretrained models: https://github.com/ali-vilab/FACM.
Authors: Joaquim Comas, Alexander Joel Vera, Xavier Vives, Eleonora De Filippi, Alexandre Pereda, Federico Sukno
Abstract: In recent years, affective computing and its applications have become a fast-growing research topic. Despite significant advancements, the lack of affective multi-modal datasets remains a major bottleneck in developing accurate emotion recognition systems. Furthermore, the use of contact-based devices during emotion elicitation often unintentionally influences the emotional experience, reducing or altering the genuine spontaneous emotional response. This limitation highlights the need for methods capable of extracting affective cues from multiple modalities without physical contact, such as remote physiological emotion recognition. To address this, we present the Contactless Affective States Through Physiological Signals Database (CAST-Phys), a novel high-quality dataset explicitly designed for multi-modal remote physiological emotion recognition using facial and physiological cues. The dataset includes diverse physiological signals, such as photoplethysmography (PPG), electrodermal activity (EDA), and respiration rate (RR), alongside high-resolution uncompressed facial video recordings, enabling the potential for remote signal recovery. Our analysis highlights the crucial role of physiological signals in realistic scenarios where facial expressions alone may not provide sufficient emotional information. Furthermore, we demonstrate the potential of remote multi-modal emotion recognition by evaluating the impact of individual and fused modalities, showcasing its effectiveness in advancing contactless emotion recognition technologies.
Authors: In\`es Hyeonsu Kim, Seokju Cho, Jahyeok Koo, Junghyun Park, Jiahui Huang, Honglak Lee, Joon-Young Lee, Seungryong Kim
Abstract: Point tracking models often struggle to generalize to real-world videos because large-scale training data is predominantly synthetic--the only source currently feasible to produce at scale. Collecting real-world annotations, however, is prohibitively expensive, as it requires tracking hundreds of points across frames. We introduce AnthroTAP, an automated pipeline that generates large-scale pseudo-labeled point tracking data from real human motion videos. Leveraging the structured complexity of human movement-non-rigid deformations, articulated motion, and frequent occlusions-AnthroTAP fits Skinned Multi-Person Linear (SMPL) models to detected humans, projects mesh vertices onto image planes, resolves occlusions via ray-casting, and filters unreliable tracks using optical flow consistency. A model trained on the AnthroTAP dataset achieves state-of-the-art performance on TAP-Vid, a challenging general-domain benchmark for tracking any point on diverse rigid and non-rigid objects (e.g., humans, animals, robots, and vehicles). Our approach outperforms recent self-supervised teacher-student models trained on vastly larger real datasets, while requiring only one day of training on 4 GPUs. AnthroTAP shows that structured human motion offers a scalable and effective source of real-world supervision for point tracking. Code and datasets will be made publicly available.
Authors: Enrico Vompa, Tanel Tammet, Mohit Vaishnav
Abstract: A challenge in advancing Visual-Language Models (VLMs) is determining whether their failures on abstract reasoning tasks, such as Bongard problems, stem from flawed perception or faulty top-down reasoning. To disentangle these factors, we introduce a diagnostic framework centered on the Linear Separability Ceiling (LSC), the performance achievable by a linear classifier on a VLM's raw visual embeddings. Applying this framework to state-of-the-art VLMs, we uncover a pervasive ``alignment gap'', where most models fail to generatively outperform the linear separability of their representations. We find that the few models surpassing this ceiling do so via two mechanisms: by further refining visual representations into a more linearly separable format or by executing non-linear decision logic. We demonstrate that this bottleneck is not a fundamental limitation but a solvable visual alignment issue. Our method augments standard next-token prediction with a contrastive objective to restructure the visual manifold into a more one-dimensionally linear geometry, improving image-to-image comparison and enabling models to significantly surpass the LSC on abstract binary classification tasks.
Authors: Tianyu Xiong, Dayi Tan, Wei Tian
Abstract: Animal pose estimation is a fundamental task in computer vision, with growing importance in ecological monitoring, behavioral analysis, and intelligent livestock management. Compared to human pose estimation, animal pose estimation is more challenging due to high interspecies morphological diversity, complex body structures, and limited annotated data. In this work, we introduce DiffPose-Animal, a novel diffusion-based framework for top-down animal pose estimation. Unlike traditional heatmap regression methods, DiffPose-Animal reformulates pose estimation as a denoising process under the generative framework of diffusion models. To enhance semantic guidance during keypoint generation, we leverage large language models (LLMs) to extract both global anatomical priors and local keypoint-wise semantics based on species-specific prompts. These textual priors are encoded and fused with image features via cross-attention modules to provide biologically meaningful constraints throughout the denoising process. Additionally, a diffusion-based keypoint decoder is designed to progressively refine pose predictions, improving robustness to occlusion and annotation sparsity. Extensive experiments on public animal pose datasets demonstrate the effectiveness and generalization capability of our method, especially under challenging scenarios with diverse species, cluttered backgrounds, and incomplete keypoints.
Authors: Zheng Qin, Ruobing Zheng, Yabing Wang, Tianqi Li, Yi Yuan, Jingdong Chen, Le Wang
Abstract: While Multimodal Large Language Models (MLLMs) show immense promise for achieving truly human-like interactions, progress is hindered by the lack of fine-grained evaluation frameworks for human-centered scenarios, encompassing both the understanding of complex human intentions and the provision of empathetic, context-aware responses. Here we introduce HumanSense, a comprehensive benchmark designed to evaluate the human-centered perception and interaction capabilities of MLLMs, with a particular focus on deep understanding of extended multimodal contexts and the formulation of rational feedback. Our evaluation reveals that leading MLLMs still have considerable room for improvement, particularly for advanced interaction-oriented tasks. Supplementing visual input with audio and text information yields substantial improvements, and Omni-modal models show advantages on these tasks.Furthermore, grounded in the observation that appropriate feedback stems from a contextual analysis of the interlocutor's needs and emotions, we posit that reasoning ability serves as the key to unlocking it. We devise a multi-stage, modality-progressive reinforcement learning approach, resulting in HumanSense-Omni-Reasoning, which substantially enhances performance on higher-level understanding and interactive tasks. Additionally, we observe that successful reasoning processes appear to exhibit consistent thought patterns. By designing corresponding prompts, we also enhance the performance of non-reasoning models in a training-free manner.Project page: \textcolor{brightpink}{https://digital-avatar.github.io/ai/HumanSense/}
Authors: Guoqing Zhang, Xingtong Ge, Lu Shi, Xin Zhang, Muqing Xue, Wanru Xu, Yigang Cen, Yidong Li
Abstract: The image-to-image generation task aims to produce controllable images by leveraging conditional inputs and prompt instructions. However, existing methods often train separate control branches for each type of condition, leading to redundant model structures and inefficient use of computational resources. To address this, we propose a Unified image-to-image Generation (UniGen) framework that supports diverse conditional inputs while enhancing generation efficiency and expressiveness. Specifically, to tackle the widely existing parameter redundancy and computational inefficiency in controllable conditional generation architectures, we propose the Condition Modulated Expert (CoMoE) module. This module aggregates semantically similar patch features and assigns them to dedicated expert modules for visual representation and conditional modeling. By enabling independent modeling of foreground features under different conditions, CoMoE effectively mitigates feature entanglement and redundant computation in multi-condition scenarios. Furthermore, to bridge the information gap between the backbone and control branches, we propose WeaveNet, a dynamic, snake-like connection mechanism that enables effective interaction between global text-level control from the backbone and fine-grained control from conditional branches. Extensive experiments on the Subjects-200K and MultiGen-20M datasets across various conditional image generation tasks demonstrate that our method consistently achieves state-of-the-art performance, validating its advantages in both versatility and effectiveness. The code has been uploaded to https://github.com/gavin-gqzhang/UniGen.
Authors: Shay Shomer-Chai, Wenxuan Peng, Bharath Hariharan, Hadar Averbuch-Elor
Abstract: Text-to-image generation has recently seen remarkable success, granting users with the ability to create high-quality images through the use of text. However, contemporary methods face challenges in capturing the precise semantics conveyed by complex multi-object prompts. Consequently, many works have sought to mitigate such semantic misalignments, typically via inference-time schemes that modify the attention layers of the denoising networks. However, prior work has mostly utilized coarse metrics, such as the cosine similarity between text and image CLIP embeddings, or human evaluations, which are challenging to conduct on a larger-scale. In this work, we perform a case study on colors -- a fundamental attribute commonly associated with objects in text prompts, which offer a rich test bed for rigorous evaluation. Our analysis reveals that pretrained models struggle to generate images that faithfully reflect multiple color attributes-far more so than with single-color prompts-and that neither inference-time techniques nor existing editing methods reliably resolve these semantic misalignments. Accordingly, we introduce a dedicated image editing technique, mitigating the issue of multi-object semantic alignment for prompts containing multiple colors. We demonstrate that our approach significantly boosts performance over a wide range of metrics, considering images generated by various text-to-image diffusion-based techniques.
Authors: Zhexi Peng, Kun Zhou, Tianjia Shao
Abstract: While recent Gaussian-based SLAM methods achieve photorealistic reconstruction from RGB-D data, their computational performance remains a critical bottleneck. State-of-the-art techniques operate at less than 20 fps, significantly lagging behind geometry-based approaches like KinectFusion (hundreds of fps). This limitation stems from the heavy computational burden: modeling scenes requires numerous Gaussians and complex iterative optimization to fit RGB-D data; insufficient Gaussian counts or optimization iterations cause severe quality degradation. To address this, we propose a Gaussian-SDF hybrid representation, combining a colorized signed distance field (SDF) for smooth geometry and appearance with 3D Gaussians to capture underrepresented details. The SDF is efficiently constructed via RGB-D fusion (as in geometry-based methods), while Gaussians undergo iterative optimization. Our representation enables significant Gaussian reduction (50% fewer) by avoiding full-scene Gaussian modeling, and efficient Gaussian optimization (75% fewer iterations) through targeted appearance refinement. Building upon this representation, we develop GPS-SLAM (Gaussian-plus-SDF SLAM), a real-time 3D reconstruction system achieving over 150 fps on real-world Azure Kinect sequences, faster by an order-of-magnitude than state-of-the-art techniques while maintaining comparable reconstruction quality. The source code and data are available at https://gapszju.github.io/GPS-SLAM.
Authors: Akwasi Asare, Thanh-Huy Nguyen, Ulas Bagci
Abstract: Colorectal cancer (CRC) remains one of the leading causes of cancer-related morbidity and mortality worldwide, with gastrointestinal (GI) polyps serving as critical precursors according to the World Health Organization (WHO). Early and accurate segmentation of polyps during colonoscopy is essential for reducing CRC progression, yet manual delineation is labor-intensive and prone to observer variability. Deep learning methods have demonstrated strong potential for automated polyp analysis, but their limited interpretability remains a barrier to clinical adoption. In this study, we present PolypSeg-GradCAM, an explainable deep learning framework that integrates a U-Net architecture with a pre-trained ResNet-34 backbone and Gradient-weighted Class Activation Mapping (Grad-CAM) for transparent polyp segmentation. To ensure rigorous benchmarking, the model was trained and evaluated using 5-Fold Cross-Validation on the Kvasir-SEG dataset of 1,000 annotated endoscopic images. Experimental results show a mean Dice coefficient of 0.8902 +/- 0.0125, a mean Intersection-over-Union (IoU) of 0.8023, and an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.9722. Advanced quantitative analysis using an optimal threshold yielded a Sensitivity of 0.9058 and Precision of 0.9083. Additionally, Grad-CAM visualizations confirmed that predictions were guided by clinically relevant regions, offering insight into the model's decision-making process. This study demonstrates that integrating segmentation accuracy with interpretability can support the development of trustworthy AI-assisted colonoscopy tools.
Authors: Miao Jing, Mengting Jia, Junling Lin, Zhongxia Shen, Huan Gao, Mingkun Xu, Shangyang Li
Abstract: Recent advances in vision-language models (VLMs) have achieved remarkable performance on standard medical benchmarks, yet their true clinical reasoning ability remains unclear. Existing datasets predominantly emphasize classification accuracy, creating an evaluation illusion in which models appear proficient while still failing at high-stakes diagnostic reasoning. We introduce Neural-MedBench, a compact yet reasoning-intensive benchmark specifically designed to probe the limits of multimodal clinical reasoning in neurology. Neural-MedBench integrates multi-sequence MRI scans, structured electronic health records, and clinical notes, and encompasses three core task families: differential diagnosis, lesion recognition, and rationale generation. To ensure reliable evaluation, we develop a hybrid scoring pipeline that combines LLM-based graders, clinician validation, and semantic similarity metrics. Through systematic evaluation of state-of-the-art VLMs, including GPT-4o, Claude-4, and MedGemma, we observe a sharp performance drop compared to conventional datasets. Error analysis shows that reasoning failures, rather than perceptual errors, dominate model shortcomings. Our findings highlight the necessity of a Two-Axis Evaluation Framework: breadth-oriented large datasets for statistical generalization, and depth-oriented, compact benchmarks such as Neural-MedBench for reasoning fidelity. We release Neural-MedBench at https://neuromedbench.github.io/ as an open and extensible diagnostic testbed, which guides the expansion of future benchmarks and enables rigorous yet cost-effective assessment of clinically trustworthy AI.
Authors: Emadeldeen Hamdan, Ahmet Enis Cetin
Abstract: Reducing the cost of multiplications is critical for efficient deep neural network deployment, especially in energy-constrained edge devices. In this work, we introduce HTMA-Net, a novel framework that integrates the Hadamard Transform (HT) with multiplication-avoiding (MA) SRAM-based in-memory computing to reduce arithmetic complexity while maintaining accuracy. Unlike prior methods that only target multiplications in convolutional layers or focus solely on in-memory acceleration, HTMA-Net selectively replaces intermediate convolutions with Hybrid Hadamard-based transform layers whose internal convolutions are implemented via multiplication-avoiding in-memory operations. We evaluate HTMA-Net on ResNet-18 using CIFAR-10, CIFAR-100, and Tiny ImageNet, and provide a detailed comparison against regular, MF-only, and HT-only variants. Results show that HTMA-Net eliminates up to 52\% of multiplications compared to baseline ResNet-18, ResNet-20, and ResNet-32 models, while achieving comparable accuracy in evaluation and significantly reducing computational complexity and the number of parameters. Our results demonstrate that combining structured Hadamard transform layers with SRAM-based in-memory computing multiplication-avoiding operators is a promising path towards efficient deep learning architectures.
Authors: Siheng Wang, Zhengdao Li, Yanshu Li, Canran Xiao, Haibo Zhan, Zhengtao Yao, Xuzhi Zhang, Jiale Kang, Linshan Li, Weiming Liu, Zhikang Dong, Jifeng Shen, Junhao Dong, Qiang Sun, Piotr Koniusz
Abstract: Object detection has advanced significantly in the closed-set setting, but real-world deployment remains limited by two challenges: poor generalization to unseen categories and insufficient robustness under adverse conditions. Prior research has explored these issues separately: visible-infrared detection improves robustness but lacks generalization, while open-world detection leverages vision-language alignment strategy for category diversity but struggles under extreme environments. This trade-off leaves robustness and diversity difficult to achieve simultaneously. To mitigate these issues, we propose \textbf{C3-OWD}, a curriculum cross-modal contrastive learning framework that unifies both strengths. Stage~1 enhances robustness by pretraining with RGBT data, while Stage~2 improves generalization via vision-language alignment. To prevent catastrophic forgetting between two stages, we introduce an Exponential Moving Average (EMA) mechanism that theoretically guarantees preservation of pre-stage performance with bounded parameter lag and function consistency. Experiments on FLIR, OV-COCO, and OV-LVIS demonstrate the effectiveness of our approach: C3-OWD achieves $80.1$ AP$^{50}$ on FLIR, $48.6$ AP$^{50}_{\text{Novel}}$ on OV-COCO, and $35.7$ mAP$_r$ on OV-LVIS, establishing competitive performance across both robustness and diversity evaluations. Code available at: https://github.com/justin-herry/C3-OWD.git.
Authors: Xiang An, Yin Xie, Kaicheng Yang, Wenkang Zhang, Xiuwei Zhao, Zheng Cheng, Yirui Wang, Songcen Xu, Changrui Chen, Didi Zhu, Chunsheng Wu, Huajie Tan, Chunyuan Li, Jing Yang, Jie Yu, Xiyao Wang, Bin Qin, Yumeng Wang, Zizhen Yan, Ziyong Feng, Ziwei Liu, Bo Li, Jiankang Deng
Abstract: We present LLaVA-OneVision-1.5, a novel family of Large Multimodal Models (LMMs) that achieve state-of-the-art performance with significantly reduced computational and financial costs. Different from the existing works, LLaVA-OneVision-1.5 provides an open, efficient, and reproducible framework for building high-quality vision-language models entirely from scratch. The LLaVA-OneVision-1.5 release comprises three primary components: (1) Large-Scale Curated Datasets: We construct an 85M concept-balanced pretraining dataset LLaVA-OneVision-1.5-Mid-Traning and a meticulously curated 22M instruction dataset LLaVA-OneVision-1.5-Instruct. (2) Efficient Training Framework: We develop a complete end-to-end efficient training framework leveraging an offline parallel data packing strategy to facilitate the training of LLaVA-OneVision-1.5 within a $16,000 budget. (3) State-of-the-art Performance: Experimental results demonstrate that LLaVA-OneVision-1.5 yields exceptionally competitive performance across a broad range of downstream tasks. Specifically, LLaVA-OneVision-1.5-8B outperforms Qwen2.5-VL-7B on 18 of 27 benchmarks, and LLaVA-OneVision-1.5-4B surpasses Qwen2.5-VL-3B on all 27 benchmarks. (4) RL-based Post-training: We unlock the model's latent potential through a lightweight RL stage, effectively eliciting robust chain-of-thought reasoning to significantly boost performance on complex multimodal reasoning tasks.
Authors: Shuai Shao, Shu Jiang, Shiyuan Zhao, Di Yang, Yan Wang, Yutong Bai, Jianguo Zhang, Jiangtao Wang
Abstract: Parkinson's disease (PD) is a common neurodegenerative disorder that severely diminishes patients' quality of life. Its global prevalence has increased markedly in recent decades. Current diagnostic workflows are complex and heavily reliant on neurologists' expertise, often resulting in delays in early detection and missed opportunities for timely intervention. To address these issues, we propose an end-to-end automated diagnostic method for PD, termed PD-Diag-Net, which performs risk assessment and auxiliary diagnosis directly from raw MRI scans. This framework first introduces an MRI Pre-processing Module (MRI-Processor) to mitigate inter-subject and inter-scanner variability by flexibly integrating established medical imaging preprocessing tools. It then incorporates two forms of clinical prior knowledge: (1) Brain-Region-Relevance-Prior (Relevance-Prior), which specifies brain regions strongly associated with PD; and (2) Brain-Region-Aging-Prior (Aging-Prior), which reflects the accelerated aging typically observed in PD-associated regions. Building on these priors, we design two dedicated modules: the Relevance-Prior Guided Feature Aggregation Module (Aggregator), which guides the model to focus on PD-associated regions at the inter-subject level, and the Age-Prior Guided Diagnosis Module (Diagnoser), which leverages brain age gaps as auxiliary constraints at the intra-subject level to enhance diagnostic accuracy and clinical interpretability. Furthermore, we collected external test data from our collaborating hospital. Experimental results show that PD-Diag-Net achieves 86\% accuracy on external tests and over 96% accuracy in early-stage diagnosis, outperforming existing advanced methods by more than 20%.
Authors: Yi Yang, Xiaokun Zhang, Qingchen Fang, Jing Liu, Ziqi Ye, Rui Li, Li Liu, Haipeng Wang
Abstract: Cross-modal artificial intelligence, represented by visual language models, has achieved significant success in general image understanding. However, a fundamental cognitive inconsistency exists between general visual representation and remote sensing image interpretation: remote sensing images couple topography, terrain, and spatial structure, thereby inherently requiring models to possess deep geoscientific understanding. This cognitive difference is further amplified in synthetic aperture radar (SAR) imagery: while SAR possesses irreplaceable all-weather, all-day observation capabilities, it is constrained by coherent imaging mechanisms, exhibiting significant modal heterogeneity with general images. To address this inconsistency, we propose FUSAR-KLIP, the first knowledge-guided general multimodal foundational model for SAR, along with reusable data and evaluation baselines. Specifically: (1) FUSAR-GEOVL-1M (the first large-scale SAR dataset with complete geographic projection attributes) was constructed, covering multiple satellite platforms, 120,000 images, and 135 cities; (2) Aligned structured text was generated through hierarchical cognitive thought chains, accurately encoding more than 1 million multidimensional semantic information from geomorphological environment and regional attributes to spatial relationships; (3) A self-consistent iterative optimization mechanism was designed to guide cross-modal learning with this knowledge information consistent with human cognition and physical laws in a self-supervised closed loop consisting of contrast, matching, and reconstruction; (4) A unified evaluation benchmark was established in 11 typical downstream tasks in the two major categories of vision and language, and compared with 15 mainstream foundation models.
Authors: Victor Sechaud, J\'er\'emy Scanvic, Quentin Barth\'elemy, Patrice Abry, Juli\'an Tachella
Abstract: Self-supervised learning for inverse problems allows to train a reconstruction network from noise and/or incomplete data alone. These methods have the potential of enabling learning-based solutions when obtaining ground-truth references for training is expensive or even impossible. In this paper, we propose a new self-supervised learning strategy devised for the challenging setting where measurements are observed via a single incomplete observation model. We introduce a new definition of equivariance in the context of reconstruction networks, and show that the combination of self-supervised splitting losses and equivariant reconstruction networks results in unbiased estimates of the supervised loss. Through a series of experiments on image inpainting, accelerated magnetic resonance imaging, sparse-view computed tomography, and compressive sensing, we demonstrate that the proposed loss achieves state-of-the-art performance in settings with highly rank-deficient forward models. The code is available at https://github.com/vsechaud/Equivariant-Splitting
Authors: Honglin Liu, Chao Sun, Peng Hu, Yunfan Li, Xi Peng
Abstract: Conventional representation learning methods learn a universal representation that primarily captures dominant semantics, which may not always align with customized downstream tasks. For instance, in animal habitat analysis, researchers prioritize scene-related features, whereas universal embeddings emphasize categorical semantics, leading to suboptimal results. As a solution, existing approaches resort to supervised fine-tuning, which however incurs high computational and annotation costs. In this paper, we propose Conditional Representation Learning (CRL), aiming to extract representations tailored to arbitrary user-specified criteria. Specifically, we reveal that the semantics of a space are determined by its basis, thereby enabling a set of descriptive words to approximate the basis for a customized feature space. Building upon this insight, given a user-specified criterion, CRL first employs a large language model (LLM) to generate descriptive texts to construct the semantic basis, then projects the image representation into this conditional feature space leveraging a vision-language model (VLM). The conditional representation better captures semantics for the specific criterion, which could be utilized for multiple customized tasks. Extensive experiments on classification and retrieval tasks demonstrate the superiority and generality of the proposed CRL. The code is available at https://github.com/XLearning-SCU/2025-NeurIPS-CRL.
Authors: Minkyoung Cho, Ruben Ohana, Christian Jacobsen, Adityan Jothi, Min-Hung Chen, Z. Morley Mao, Ethem Can
Abstract: Current controllable diffusion models typically rely on fixed architectures that modify intermediate activations to inject guidance conditioned on a new modality. This approach uses a static conditioning strategy for a dynamic, multi-stage denoising process, limiting the model's ability to adapt its response as the generation evolves from coarse structure to fine detail. We introduce TC-LoRA (Temporally Modulated Conditional LoRA), a new paradigm that enables dynamic, context-aware control by conditioning the model's weights directly. Our framework uses a hypernetwork to generate LoRA adapters on-the-fly, tailoring weight modifications for the frozen backbone at each diffusion step based on time and the user's condition. This mechanism enables the model to learn and execute an explicit, adaptive strategy for applying conditional guidance throughout the entire generation process. Through experiments on various data domains, we demonstrate that this dynamic, parametric control significantly enhances generative fidelity and adherence to spatial conditions compared to static, activation-based methods. TC-LoRA establishes an alternative approach in which the model's conditioning strategy is modified through a deeper functional adaptation of its weights, allowing control to align with the dynamic demands of the task and generative stage.
Authors: Lishen Qu, Zhihao Liu, Jinshan Pan, Shihao Zhou, Jinglei Shi, Duosheng Chen, Jufeng Yang
Abstract: Lens flare occurs when shooting towards strong light sources, significantly degrading the visual quality of images. Due to the difficulty in capturing flare-corrupted and flare-free image pairs in the real world, existing datasets are typically synthesized in 2D by overlaying artificial flare templates onto background images. However, the lack of flare diversity in templates and the neglect of physical principles in the synthesis process hinder models trained on these datasets from generalizing well to real-world scenarios. To address these challenges, we propose a new physics-informed method for flare data generation, which consists of three stages: parameterized template creation, the laws of illumination-aware 2D synthesis, and physical engine-based 3D rendering, which finally gives us a mixed flare dataset that incorporates both 2D and 3D perspectives, namely FlareX. This dataset offers 9,500 2D templates derived from 95 flare patterns and 3,000 flare image pairs rendered from 60 3D scenes. Furthermore, we design a masking approach to obtain real-world flare-free images from their corrupted counterparts to measure the performance of the model on real-world images. Extensive experiments demonstrate the effectiveness of our method and dataset.
Authors: Long Cui, Weiyun Wang, Jie Shao, Zichen Wen, Gen Luo, Linfeng Zhang, Yanting Zhang, Yu Qiao, Wenhai Wang
Abstract: Existing Multimodal Large Language Models (MLLMs) suffer from increased inference costs due to the additional vision tokens introduced by image inputs. In this work, we propose Visual Consistency Learning (ViCO), a novel training algorithm that enables the model to represent images of varying semantic complexities using different numbers of vision tokens. The key idea behind our method is to employ multiple MLP connectors, each with a different image compression ratio, to downsample the vision tokens based on the semantic complexity of the image. During training, we minimize the KL divergence between the responses conditioned on different MLP connectors. At inference time, we introduce an image router, termed Visual Resolution Router (ViR), that automatically selects the appropriate compression rate for each image patch. Compared with existing dynamic high-resolution strategies, which adjust the number of visual tokens based on image resolutions, our method dynamically adapts the number of visual tokens according to semantic complexity. Experimental results demonstrate that our method can reduce the number of vision tokens by up to 50% while maintaining the model's perception, reasoning, and OCR capabilities. We hope this work will contribute to the development of more efficient MLLMs. The code and models will be released to facilitate future research.
Authors: Shuai Fu, Jian Zhou, Qi Chen, Huang Jing, Huy Anh Nguyen, Xiaohan Liu, Zhixiong Zeng, Lin Ma, Quanshi Zhang, Qi Wu
Abstract: Diffusion probabilistic models (DPMs) have demonstrated remarkable progress in generative tasks, such as image and video synthesis. However, they still often produce hallucinated samples (hallucinations) that conflict with real-world knowledge, such as generating an implausible duplicate cup floating beside another cup. Despite their prevalence, the lack of feasible methodologies for systematically quantifying such hallucinations hinders progress in addressing this challenge and obscures potential pathways for designing next-generation generative models under factual constraints. In this work, we bridge this gap by focusing on a specific form of hallucination, which we term counting hallucination, referring to the generation of an incorrect number of instances or structured objects, such as a hand image with six fingers, despite such patterns being absent from the training data. To this end, we construct a dataset suite CountHalluSet, with well-defined counting criteria, comprising ToyShape, SimObject, and RealHand. Using these datasets, we develop a standardized evaluation protocol for quantifying counting hallucinations, and systematically examine how different sampling conditions in DPMs, including solver type, ODE solver order, sampling steps, and initial noise, affect counting hallucination levels. Furthermore, we analyze their correlation with common evaluation metrics such as FID, revealing that this widely used image quality metric fails to capture counting hallucinations consistently. This work aims to take the first step toward systematically quantifying hallucinations in diffusion models and offer new insights into the investigation of hallucination phenomena in image generation.
Authors: JoungBin Lee, Jaewoo Jung, Jisang Han, Takuya Narihira, Kazumi Fukuda, Junyoung Seo, Sunghwan Hong, Yuki Mitsufuji, Seungryong Kim
Abstract: We present 3DScenePrompt, a framework that generates the next video chunk from arbitrary-length input while enabling precise camera control and preserving scene consistency. Unlike methods conditioned on a single image or a short clip, we employ dual spatio-temporal conditioning that reformulates context-view referencing across the input video. Our approach conditions on both temporally adjacent frames for motion continuity and spatially adjacent content for scene consistency. However, when generating beyond temporal boundaries, directly using spatially adjacent frames would incorrectly preserve dynamic elements from the past. We address this by introducing a 3D scene memory that represents exclusively the static geometry extracted from the entire input video. To construct this memory, we leverage dynamic SLAM with our newly introduced dynamic masking strategy that explicitly separates static scene geometry from moving elements. The static scene representation can then be projected to any target viewpoint, providing geometrically consistent warped views that serve as strong 3D spatial prompts while allowing dynamic regions to evolve naturally from temporal context. This enables our model to maintain long-range spatial coherence and precise camera control without sacrificing computational efficiency or motion realism. Extensive experiments demonstrate that our framework significantly outperforms existing methods in scene consistency, camera controllability, and generation quality. Project page : https://cvlab-kaist.github.io/3DScenePrompt/
Authors: Guangyu Dai, Dong Chen, Siliang Tang, Yueting Zhuang
Abstract: Video anomaly detection (VAD) is a challenging task that detects anomalous frames in continuous surveillance videos. Most previous work utilizes the spatio-temporal correlation of visual features to distinguish whether there are abnormalities in video snippets. Recently, some works attempt to introduce multi-modal information, like text feature, to enhance the results of video anomaly detection. However, these works merely incorporate text features into video snippets in a coarse manner, overlooking the significant amount of redundant information that may exist within the video snippets. Therefore, we propose to leverage the diversity among multi-modal information to further refine the extracted features, reducing the redundancy in visual features, and we propose Grained Multi-modal Feature for Video Anomaly Detection (GMFVAD). Specifically, we generate more grained multi-modal feature based on the video snippet, which summarizes the main content, and text features based on the captions of original video will be introduced to further enhance the visual features of highlighted portions. Experiments show that the proposed GMFVAD achieves state-of-the-art performance on four mainly datasets. Ablation experiments also validate that the improvement of GMFVAD is due to the reduction of redundant information.
Authors: Bingchen Miao, Rong Wei, Zhiqi Ge, Xiaoquan sun, Shiqi Gao, Jingzhe Zhu, Renhan Wang, Siliang Tang, Jun Xiao, Rui Tang, Juncheng Li
Abstract: 3D Gaussian Splatting (3DGS), a 3D representation method with photorealistic real-time rendering capabilities, is regarded as an effective tool for narrowing the sim-to-real gap. However, it lacks fine-grained semantics and physical executability for Visual-Language Navigation (VLN). To address this, we propose SAGE-3D (Semantically and Physically Aligned Gaussian Environments for 3D Navigation), a new paradigm that upgrades 3DGS into an executable, semantically and physically aligned environment. It comprises two components: (1) Object-Centric Semantic Grounding, which adds object-level fine-grained annotations to 3DGS; and (2) Physics-Aware Execution Jointing, which embeds collision objects into 3DGS and constructs rich physical interfaces. We release InteriorGS, containing 1K object-annotated 3DGS indoor scene data, and introduce SAGE-Bench, the first 3DGS-based VLN benchmark with 2M VLN data. Experiments show that 3DGS scene data is more difficult to converge, while exhibiting strong generalizability, improving baseline performance by 31% on the VLN-CE Unseen task. Our data and code are available at: https://sage-3d.github.io.
Authors: Bill Psomas, George Retsinas, Nikos Efthymiadis, Panagiotis Filntisis, Yannis Avrithis, Petros Maragos, Ondrej Chum, Giorgos Tolias
Abstract: The progress of composed image retrieval (CIR), a popular research direction in image retrieval, where a combined visual and textual query is used, is held back by the absence of high-quality training and evaluation data. We introduce a new evaluation dataset, i-CIR, which, unlike existing datasets, focuses on an instance-level class definition. The goal is to retrieve images that contain the same particular object as the visual query, presented under a variety of modifications defined by textual queries. Its design and curation process keep the dataset compact to facilitate future research, while maintaining its challenge-comparable to retrieval among more than 40M random distractors-through a semi-automated selection of hard negatives. To overcome the challenge of obtaining clean, diverse, and suitable training data, we leverage pre-trained vision-and-language models (VLMs) in a training-free approach called BASIC. The method separately estimates query-image-to-image and query-text-to-image similarities, performing late fusion to upweight images that satisfy both queries, while down-weighting those that exhibit high similarity with only one of the two. Each individual similarity is further improved by a set of components that are simple and intuitive. BASIC sets a new state of the art on i-CIR but also on existing CIR datasets that follow a semantic-level class definition. Project page: https://vrg.fel.cvut.cz/icir/.
Authors: Yinglu Li, Zhiying Lu, Zhihang Liu, Chuanbin Liu, Hongtao Xie
Abstract: Multi-modal Retrieval-Augmented Generation (RAG) has become a critical method for empowering LLMs by leveraging candidate visual documents. However, current methods consider the entire document as the basic retrieval unit, introducing substantial irrelevant visual content in two ways: 1) Relevant documents often contain large regions unrelated to the query, diluting the focus on salient information; 2) Retrieving multiple documents to increase recall further introduces redundant and irrelevant documents. These redundant contexts distract the model's attention and further degrade the performance. To address this challenge, we propose RegionRAG, a novel framework that shifts the retrieval paradigm from the document level to the region level. During training, we design a hybrid supervision strategy from both labeled data and unlabeled data to pinpoint relevant patches. During inference, we propose a dynamic pipeline that intelligently groups salient patches into complete semantic regions. By delegating the task of identifying relevant regions to the retriever, RegionRAG enables the generator to focus solely on concise, query-relevant visual content, improving both efficiency and accuracy. Experiments on six benchmarks demonstrate that RegionRAG achieves state-of-the-art performance. It improves retrieval accuracy by 10.02% in R@1 on average, and boosts question answering accuracy by 3.56% while using only 71.42% visual tokens compared with prior methods.
Authors: Panwang Pan, Jingjing Zhao, Yuchen Lin, Chenguo Lin, Chenxin Li, Hengyu Liu, Tingting Shen, Yadong MU
Abstract: Significant progress has been achieved in high-fidelity video synthesis, yet current paradigms often fall short in effectively integrating identity information from multiple subjects. This leads to semantic conflicts and suboptimal performance in preserving identities and interactions, limiting controllability and applicability. To tackle this issue, we introduce ID-Crafter, a framework for multi-subject video generation that achieves superior identity preservation and semantic coherence. ID-Crafter integrates three key components: (i) a hierarchical identity-preserving attention mechanism that progressively aggregates features at intra-subject, inter-subject, and cross-modal levels; (ii) a semantic understanding module powered by a pretrained Vision-Language Model (VLM) to provide fine-grained guidance and capture complex inter-subject relationships; and (iii) an online reinforcement learning phase to further refine the model for critical concepts. Furthermore, we construct a new dataset to facilitate robust training and evaluation. Extensive experiments demonstrate that ID-Crafter establishes new state-of-the-art performance on multi-subject video generation benchmarks, excelling in identity preservation, temporal consistency, and overall video quality. Project page: https://angericky.github.io/ID-Crafter
Authors: Thomas Manzini, Priyankari Perali, Robin R. Murphy
Abstract: This paper presents the first AI/ML system for automating building damage assessment in uncrewed aerial systems (sUAS) imagery to be deployed operationally during federally declared disasters (Hurricanes Debby and Helene). In response to major disasters, sUAS teams are dispatched to collect imagery of the affected areas to assess damage; however, at recent disasters, teams collectively delivered between 47GB and 369GB of imagery per day, representing more imagery than can reasonably be transmitted or interpreted by subject matter experts in the disaster scene, thus delaying response efforts. To alleviate this data avalanche encountered in practice, computer vision and machine learning techniques are necessary. While prior work has been deployed to automatically assess damage in satellite imagery, there is no current state of practice for sUAS-based damage assessment systems, as all known work has been confined to academic settings. This work establishes the state of practice via the development and deployment of models for building damage assessment with sUAS imagery. The model development involved training on the largest known dataset of post-disaster sUAS aerial imagery, containing 21,716 building damage labels, and the operational training of 91 disaster practitioners. The best performing model was deployed during the responses to Hurricanes Debby and Helene, where it assessed a combined 415 buildings in approximately 18 minutes. This work contributes documentation of the actual use of AI/ML for damage assessment during a disaster and lessons learned to the benefit of the AI/ML research and user communities.
Authors: Seyedehanita Madani, Rama Chellappa, Vishal M. Patel
Abstract: Change detection (CD) is fundamental to computer vision and remote sensing, supporting applications in environmental monitoring, disaster response, and urban development. Most CD models assume co-registered inputs, yet real-world imagery often exhibits parallax, viewpoint shifts, and long temporal gaps that cause severe misalignment. Traditional two stage methods that first register and then detect, as well as recent joint frameworks (e.g., BiFA, ChangeRD), still struggle under large displacements, relying on regression only flow, global homographies, or synthetic perturbations. We present DiffRegCD, an integrated framework that unifies dense registration and change detection in a single model. DiffRegCD reformulates correspondence estimation as a Gaussian smoothed classification task, achieving sub-pixel accuracy and stable training. It leverages frozen multi-scale features from a pretrained denoising diffusion model, ensuring robustness to illumination and viewpoint variation. Supervision is provided through controlled affine perturbations applied to standard CD datasets, yielding paired ground truth for both flow and change detection without pseudo labels. Extensive experiments on aerial (LEVIR-CD, DSIFN-CD, WHU-CD, SYSU-CD) and ground level (VL-CMU-CD) datasets show that DiffRegCD consistently surpasses recent baselines and remains reliable under wide temporal and geometric variation, establishing diffusion features and classification based correspondence as a strong foundation for unified change detection.
Authors: Yifei Sun, Yuzhi He, Junhao Jia, Jinhong Wang, Ruiquan Ge, Changmiao Wang, Hongxia Xu
Abstract: Microaneurysms (MAs), the earliest pathognomonic signs of Diabetic Retinopathy (DR), present as sub-60 $\mu m$ lesions in fundus images with highly variable photometric and morphological characteristics, rendering manual screening not only labor-intensive but inherently error-prone. While diffusion-based anomaly detection has emerged as a promising approach for automated MA screening, its clinical application is hindered by three fundamental limitations. First, these models often fall prey to "identity mapping", where they inadvertently replicate the input image. Second, they struggle to distinguish MAs from other anomalies, leading to high false positives. Third, their suboptimal reconstruction of normal features hampers overall performance. To address these challenges, we propose a Wavelet Diffusion Transformer framework for MA Detection (WDT-MD), which features three key innovations: a noise-encoded image conditioning mechanism to avoid "identity mapping" by perturbing image conditions during training; pseudo-normal pattern synthesis via inpainting to introduce pixel-level supervision, enabling discrimination between MAs and other anomalies; and a wavelet diffusion Transformer architecture that combines the global modeling capability of diffusion Transformers with multi-scale wavelet analysis to enhance reconstruction of normal retinal features. Comprehensive experiments on the IDRiD and e-ophtha MA datasets demonstrate that WDT-MD outperforms state-of-the-art methods in both pixel-level and image-level MA detection. This advancement holds significant promise for improving early DR screening.
Authors: Haiyi Li, Qi Chen, Denis Kalkofen, Hsiang-Ting Chen
Abstract: Recent advances in 3D Gaussian Splatting (3DGS) have achieved state-of-the-art results for novel view synthesis. However, efficiently capturing high-fidelity reconstructions of specific objects within complex scenes remains a significant challenge. A key limitation of existing active reconstruction methods is their reliance on scene-level uncertainty metrics, which are often biased by irrelevant background clutter and lead to inefficient view selection for object-centric tasks. We present OUGS, a novel framework that addresses this challenge with a more principled, physically-grounded uncertainty formulation for 3DGS. Our core innovation is to derive uncertainty directly from the explicit physical parameters of the 3D Gaussian primitives (e.g., position, scale, rotation). By propagating the covariance of these parameters through the rendering Jacobian, we establish a highly interpretable uncertainty model. This foundation allows us to then seamlessly integrate semantic segmentation masks to produce a targeted, object-aware uncertainty score that effectively disentangles the object from its environment. This allows for a more effective active view selection strategy that prioritizes views critical to improving object fidelity. Experimental evaluations on public datasets demonstrate that our approach significantly improves the efficiency of the 3DGS reconstruction process and achieves higher quality for targeted objects compared to existing state-of-the-art methods, while also serving as a robust uncertainty estimator for the global scene.
Authors: Siddeshwar Raghavan, Jiangpeng He, Fengqing Zhu
Abstract: Exemplar-Free Continual Learning (EFCL) restricts the storage of previous task data and is highly susceptible to catastrophic forgetting. While pre-trained models (PTMs) are increasingly leveraged for EFCL, existing methods often overlook the inherent imbalance of real-world data distributions. We discovered that real-world data streams commonly exhibit dual-level imbalances, dataset-level distributions combined with extreme or reversed skews within individual tasks, creating both intra-task and inter-task disparities that hinder effective learning and generalization. To address these challenges, we propose PANDA, a Patch-and-Distribution-Aware Augmentation framework that integrates seamlessly with existing PTM-based EFCL methods. PANDA amplifies low-frequency classes by using a CLIP encoder to identify representative regions and transplanting those into frequent-class samples within each task. Furthermore, PANDA incorporates an adaptive balancing strategy that leverages prior task distributions to smooth inter-task imbalances, reducing the overall gap between average samples across tasks and enabling fairer learning with frozen PTMs. Extensive experiments and ablation studies demonstrate PANDA's capability to work with existing PTM-based CL methods, improving accuracy and reducing catastrophic forgetting.
Authors: Dexin Zuo, Ang Li, Wei Wang, Wenxian Yu, Danping Zou
Abstract: Object 6D pose estimation, a crucial task for robotics and augmented reality applications, becomes particularly challenging when dealing with novel objects whose 3D models are not readily available. To reduce dependency on 3D models, recent studies have explored one-reference-based pose estimation, which requires only a single reference view instead of a complete 3D model. However, existing methods that rely on real-valued coordinate regression suffer from limited global consistency due to the local nature of convolutional architectures and face challenges in symmetric or occluded scenarios owing to a lack of uncertainty modeling. We present CoordAR, a novel autoregressive framework for one-reference 6D pose estimation of unseen objects. CoordAR formulates 3D-3D correspondences between the reference and query views as a map of discrete tokens, which is obtained in an autoregressive and probabilistic manner. To enable accurate correspondence regression, CoordAR introduces 1) a novel coordinate map tokenization that enables probabilistic prediction over discretized 3D space; 2) a modality-decoupled encoding strategy that separately encodes RGB appearance and coordinate cues; and 3) an autoregressive transformer decoder conditioned on both position-aligned query features and the partially generated token sequence. With these novel mechanisms, CoordAR significantly outperforms existing methods on multiple benchmarks and demonstrates strong robustness to symmetry, occlusion, and other challenges in real-world tests.
Authors: Yu Zhu, Dan Zeng, Shuiwang Li, Qijun Zhao, Qiaomu Shen, Bo Tang
Abstract: Recent research in Category-Agnostic Pose Estimation (CAPE) has adopted fixed textual keypoint description as semantic prior for two-stage pose matching frameworks. While this paradigm enhances robustness and flexibility by disentangling the dependency of support images, our critical analysis reveals two inherent limitations of static joint embedding: (1) polysemy-induced cross-category ambiguity during the matching process(e.g., the concept "leg" exhibiting divergent visual manifestations across humans and furniture), and (2) insufficient discriminability for fine-grained intra-category variations (e.g., posture and fur discrepancies between a sleeping white cat and a standing black cat). To overcome these challenges, we propose a new framework that innovatively integrates hierarchical cross-modal interaction with dual-stream feature refinement, enhancing the joint embedding with both class-level and instance-specific cues from textual description and specific images. Experiments on the MP-100 dataset demonstrate that, regardless of the network backbone, CapeNext consistently outperforms state-of-the-art CAPE methods by a large margin.
Authors: Hanzhe Liang, Jie Zhou, Can Gao, Bingyang Guo, Jinbao Wang, Linlin Shen
Abstract: 3D anomaly detection (AD) is a crucial task in computer vision, aiming to identify anomalous points or regions from point cloud data. However, existing methods may encounter challenges when handling point clouds with changes in orientation and position because the resulting features may vary significantly. To address this problem, we propose a novel Rotationally Invariant Features (RIF) framework for 3D AD. Firstly, to remove the adverse effect of variations on point cloud data, we develop a Point Coordinate Mapping (PCM) technique, which maps each point into a rotationally invariant space to maintain consistency of representation. Then, to learn robust and discriminative features, we design a lightweight Convolutional Transform Feature Network (CTF-Net) to extract rotationally invariant features for the memory bank. To improve the ability of the feature extractor, we introduce the idea of transfer learning to pre-train the feature extractor with 3D data augmentation. Experimental results show that the proposed method achieves the advanced performance on the Anomaly-ShapeNet dataset, with an average P-AUROC improvement of 17.7\%, and also gains the best performance on the Real3D-AD dataset, with an average P-AUROC improvement of 1.6\%. The strong generalization ability of RIF has been verified by combining it with traditional feature extraction methods on anomaly detection tasks, demonstrating great potential for industrial applications.
Authors: Dingkun Zhou, Patrick P. K. Chan, Hengxu Wu, Shikang Zheng, Ruiqi Huang, Yuanjie Zhao
Abstract: Deep neural networks used for human detection are highly vulnerable to adversarial manipulation, creating safety and privacy risks in real surveillance environments. Wearable attacks offer a realistic threat model, yet existing approaches usually optimize textures frame by frame and therefore fail to maintain concealment across long video sequences with motion, pose changes, and garment deformation. In this work, a sequence-level optimization framework is introduced to generate natural, printable adversarial textures for shirts, trousers, and hats that remain effective throughout entire walking videos in both digital and physical settings. Product images are first mapped to UV space and converted into a compact palette and control-point parameterization, with ICC locking to keep all colors printable. A physically based human-garment pipeline is then employed to simulate motion, multi-angle camera viewpoints, cloth dynamics, and illumination variation. An expectation-over-transformation objective with temporal weighting is used to optimize the control points so that detection confidence is minimized across whole sequences. Extensive experiments demonstrate strong and stable concealment, high robustness to viewpoint changes, and superior cross-model transferability. Physical garments produced with sublimation printing achieve reliable suppression under indoor and outdoor recordings, confirming real-world feasibility.
Authors: Mario Markov (INSAIT, Sofia University), Stefan Maria Ailuro (INSAIT, Sofia University), Luc Van Gool (INSAIT, Sofia University), Konrad Schindler (ETH Zurich), Danda Pani Paudel (INSAIT, Sofia University)
Abstract: Predicting wildfire risk is a reasoning-intensive spatial problem that requires the integration of visual, climatic, and geographic factors to infer continuous risk maps. Existing methods lack the causal reasoning and multimodal understanding required for reliable generalization. We introduce $\textbf{FireScope-Bench}$, a large-scale dataset and benchmark that couples Sentinel-2 imagery and climate data with expert-defined risk rasters across the USA, and real wildfire events in Europe for cross-continental evaluation. Building on this dataset, we propose $\textbf{FireScope}$, a VLM-based reasoning-to-generation framework that learns from both reinforcement learning and visual supervision to predict risk rasters with complementary reasoning traces. When trained in the USA and tested in Europe, $\textbf{FireScope}$ achieves substantial performance gains, while expert feedback and automated analysis confirm that its reasoning traces are faithful and semantically meaningful. Our findings demonstrate that reasoning can ground raster prediction models, improving both generalization and interpretability. To our knowledge, this is the first framework to (1) demonstrate that language-based reasoning can improve generalization in visual generation, (2) propose a high-resolution wildfire risk model that can be applied across continents, and (3) enable systematic studies of robust cross-continental generalization for multimodal fire risk models. We believe that $\textbf{FireScope-Bench}$ has the potential to serve as a foundation for advancing reasoning-driven, interpretable and generalizable spatial modeling. Data and source code will be made publicly available.
Authors: Linxiang Su, Andr\'as Balogh
Abstract: Despite its remarkable success in zero-shot image-text matching, CLIP remains highly vulnerable to adversarial perturbations on images. As adversarial fine-tuning is prohibitively costly, recent works explore various test-time defense strategies; however, these approaches still exhibit limited robustness. In this work, we revisit this problem and propose a simple yet effective strategy: Augmentation-based Test-time Adversarial Correction (ATAC). Our method operates directly in the embedding space of CLIP, calculating augmentation-induced drift vectors to infer a semantic recovery direction and correcting the embedding based on the angular consistency of these latent drifts. Across a wide range of benchmarks, ATAC consistently achieves remarkably high robustness, surpassing that of previous state-of-the-art methods by nearly 50\% on average, all while requiring minimal computational overhead. Furthermore, ATAC retains state-of-the-art robustness in unconventional and extreme settings and even achieves nontrivial robustness against adaptive attacks. Our results demonstrate that ATAC is an efficient method in a novel paradigm for test-time adversarial defenses in the embedding space of CLIP.
Authors: Xiao Cui, Yulei Qin, Xinyue Li, Wengang Zhou, Hongsheng Li, Houqiang Li
Abstract: Dataset distillation creates a small distilled set that enables efficient training by capturing key information from the full dataset. While existing dataset distillation methods perform well on balanced datasets, they struggle under long-tailed distributions, where imbalanced class frequencies induce biased model representations and corrupt statistical estimates such as Batch Normalization (BN) statistics. In this paper, we rethink long-tailed dataset distillation by revisiting the limitations of trajectory-based methods, and instead adopt the statistical alignment perspective to jointly mitigate model bias and restore fair supervision. To this end, we introduce three dedicated components that enable unbiased recovery of distilled images and soft relabeling: (1) enhancing expert models (an observer model for recovery and a teacher model for relabeling) to enable reliable statistics estimation and soft-label generation; (2) recalibrating BN statistics via a full forward pass with dynamically adjusted momentum to reduce representation skew; (3) initializing synthetic images by incrementally selecting high-confidence and diverse augmentations via a multi-round mechanism that promotes coverage and diversity. Extensive experiments on four long-tailed benchmarks show consistent improvements over state-of-the-art methods across varying degrees of class imbalance. Notably, our approach improves top-1 accuracy by 15.6% on CIFAR-100-LT and 11.8% on Tiny-ImageNet-LT under IPC=10 and IF=10. Codes are available at https://github.com/2018cx/RLDD.
Authors: Wentao Qu, Guofeng Mei, Yang Wu, Yongshun Gong, Xiaoshui Huang, Liang Xiao
Abstract: Text-to-LiDAR generation can customize 3D data with rich structures and diverse scenes for downstream tasks. However, the scarcity of Text-LiDAR pairs often causes insufficient training priors, generating overly smooth 3D scenes. Moreover, low-quality text descriptions may degrade generation quality and controllability. In this paper, we propose a Text-to-LiDAR Diffusion Model for scene generation, named T2LDM, with a Self-Conditioned Representation Guidance (SCRG). Specifically, SCRG, by aligning to the real representations, provides the soft supervision with reconstruction details for the Denoising Network (DN) in training, while decoupled in inference. In this way, T2LDM can perceive rich geometric structures from data distribution, generating detailed objects in scenes. Meanwhile, we construct a content-composable Text-LiDAR benchmark, T2nuScenes, along with a controllability metric. Based on this, we analyze the effects of different text prompts for LiDAR generation quality and controllability, providing practical prompt paradigms and insights. Furthermore, a directional position prior is designed to mitigate street distortion, further improving scene fidelity. Additionally, by learning a conditional encoder via frozen DN, T2LDM can support multiple conditional tasks, including Sparse-to-Dense, Dense-to-Sparse, and Semantic-to-LiDAR generation. Extensive experiments in unconditional and conditional generation demonstrate that T2LDM outperforms existing methods, achieving state-of-the-art scene generation.
Authors: Nisarg K. Trivedi, Vinayak A. Belludi, Li-Yun Wang, Pardis Taghavi, Dante Lok
Abstract: Reliable depth estimation under real optical conditions remains a core challenge for camera vision in systems such as autonomous robotics and augmented reality. Despite recent progress in depth estimation and depth-of-field rendering, research remains constrained by the lack of large-scale, high-fidelity, real stereo DSLR datasets, limiting real-world generalization and evaluation of models trained on synthetic data as shown extensively in literature. We present the first high-resolution (5472$\times$3648px) stereo DSLR dataset with 18000 images, systematically varying focal length and aperture across complex real scenes and capturing the optical realism and complexity of professional camera systems. For 9 scenes with varying scene complexity, lighting and background, images are captured with two identical camera assemblies at 10 focal lengths (28-70mm) and 5 apertures (f/2.8-f/22), spanning 50 optical configurations in 2000 images per scene. This full-range optics coverage enables controlled analysis of geometric and optical effects for monocular and stereo depth estimation, shallow depth-of-field rendering, deblurring, 3D scene reconstruction and novel view synthesis. Each focal configuration has a dedicated calibration image set, supporting evaluation of classical and learning based methods for intrinsic and extrinsic calibration. The dataset features challenging visual elements such as multi-scale optical illusions, reflective surfaces, mirrors, transparent glass walls, fine-grained details, and natural / artificial ambient light variations. This work attempts to bridge the realism gap between synthetic training data and real camera optics, and demonstrates challenges with the current state-of-the-art monocular, stereo depth and depth-of-field methods. We release the dataset, calibration files, and evaluation code to support reproducible research on real-world optical generalization.
Authors: Finlay G. C. Hudson, James A. D. Gardner, William A. P. Smith
Abstract: Humans excel at constructing panoramic mental models of their surroundings, maintaining object permanence and inferring scene structure beyond visible regions. In contrast, current artificial vision systems struggle with persistent, panoramic understanding, often processing scenes egocentrically on a frame-by-frame basis. This limitation is pronounced in the Track Any Point (TAP) task, where existing methods fail to track 2D points outside the field of view. To address this, we introduce TAPVid-360, a novel task that requires predicting the 3D direction to queried scene points across a video sequence, even when far outside the narrow field of view of the observed video. This task fosters learning allocentric scene representations without needing dynamic 4D ground truth scene models for training. Instead, we exploit 360 videos as a source of supervision, resampling them into narrow field-of-view perspectives while computing ground truth directions by tracking points across the full panorama using a 2D pipeline. We introduce a new dataset and benchmark, TAPVid360-10k comprising 10k perspective videos with ground truth directional point tracking. Our baseline adapts CoTracker v3 to predict per-point rotations for direction updates, outperforming existing TAP and TAPVid 3D methods. Project page: https://finlay-hudson.github.io/tapvid360
Authors: Yumeng Li, Guang Yang, Hao Liu, Bowen Wang, Colin Zhang
Abstract: Document Layout Parsing serves as a critical gateway for Artificial Intelligence (AI) to access and interpret the world's vast stores of structured knowledge. This process,which encompasses layout detection, text recognition, and relational understanding, is particularly crucial for empowering next-generation Vision-Language Models. Current methods, however, rely on fragmented, multi-stage pipelines that suffer from error propagation and fail to leverage the synergies of joint training. In this paper, we introduce $\text{dots.ocr}$, a single Vision-Language Model that, for the first time, demonstrates the advantages of jointly learning three core tasks within a unified, end-to-end framework. This is made possible by a highly scalable data engine that synthesizes a vast multilingual corpus, empowering the model to deliver robust performance across a wide array of tasks, encompassing diverse languages, layouts, and domains. The efficacy of our unified paradigm is validated by state-of-the-art performance on the comprehensive OmniDocBench. Furthermore, to catalyze research in global document intelligence, we introduce XDocParse, a challenging new benchmark spanning 126 languages. On this testbed, $\text{dots.ocr}$ establishes a powerful new baseline, outperforming the next-best competitor by a remarkable +7.4 point margin and proving its unparalleled multilingual capabilities.
Authors: Dong Li, Jiahao Xiong, Yingda Huang, Le Chang
Abstract: We introduce PoreTrack3D, the first benchmark for dynamic 3D Gaussian splatting in pore-scale, non-rigid 3D facial trajectory tracking. It contains over 440,000 facial trajectories in total, among which more than 52,000 are longer than 10 frames, including 68 manually reviewed trajectories that span the entire 150 frames. To the best of our knowledge, PoreTrack3D is the first benchmark dataset to capture both traditional facial landmarks and pore-scale keypoints trajectory, advancing the study of fine-grained facial expressions through the analysis of subtle skin-surface motion. We systematically evaluate state-of-the-art dynamic 3D Gaussian splatting methods on PoreTrack3D, establishing the first performance baseline in this domain. Overall, the pipeline developed for this benchmark dataset's creation establishes a new framework for high-fidelity facial motion capture and dynamic 3D reconstruction. Our dataset are publicly available at: https://github.com/JHXion9/PoreTrack3D
Authors: Athos Georgiou
Abstract: Late-interaction multimodal retrieval models like ColPali achieve state-of-the-art document retrieval by embedding pages as images and computing fine-grained similarity between query tokens and visual patches. However, they return entire pages rather than specific regions, limiting utility for retrieval-augmented generation (RAG) where precise context is paramount. Conversely, OCR-based systems extract structured text with bounding box coordinates but lack semantic grounding for relevance assessment. We propose a hybrid architecture that unifies these paradigms: using ColPali's patch-level similarity scores as spatial relevance filters over OCR-extracted regions. We formalize the coordinate mapping between vision transformer patch grids and OCR bounding boxes, introduce intersection metrics for relevance propagation, and establish theoretical bounds on area efficiency. We evaluate on BBox-DocVQA with ground-truth bounding boxes. For within-page localization (given correct page retrieval), ColQwen3-4B with percentile-50 thresholding achieves 59.7% hit rate at IoU@0.5 (84.4% at IoU@0.25, 35.8% at IoU@0.7), with mean IoU of 0.569, compared to ~6.7% for random region selection. Our approach reduces context tokens by 28.8% compared to returning all OCR regions and by 52.3% compared to full-page image tokens. Our approach operates at inference time without additional training. We release Snappy, an open-source implementation at https://github.com/athrael-soju/Snappy.
Authors: Zhenkai Wu, Xiaowen Ma, Zhenliang Ni, Dengming Zhang, Han Shu, Xin Jiang, Xinghao Chen
Abstract: Vision-language models (VLMs) excel at image understanding tasks, but the large number of visual tokens imposes significant computational costs, hindering deployment on mobile devices. Many pruning methods rely solely on token importance and thus overlook inter-token redundancy, retaining numerous duplicated tokens and wasting capacity. Although some redundancy-aware approaches have been proposed, they often ignore the spatial relationships among visual tokens. This can lead to overly sparse selections of retained tokens that fail to adequately cover the regions of target objects. To address these limitations, we propose VLM-Pruner, a training-free token pruning algorithm that explicitly balances redundancy and spatial sparsity. We introduce a centrifugal token pruning paradigm that enables near-to-far selection while prioritizing the preservation of fine-grained object details. Moreover, we design a Buffering for Spatial Sparsity (BSS) criterion that defers the selection of spatially distant tokens. We further adopt a parallel greedy strategy to conduct token selection efficiently. To mitigate information loss from pruning, we selectively fuse salient information from the discarded tokens into the retained ones. Comprehensive comparisons demonstrate that VLM-Pruner consistently outperforms strong baselines across five VLMs with an 88.9\% pruning rate, while delivering an end-to-end inference speedup. The code is available at https://github.com/Casey-bit/VLMPruner.
Authors: Zhiwei Chen, Yupeng Hu, Zixu Li, Zhiheng Fu, Haokun Wen, Weili Guan
Abstract: Composed Video Retrieval (CVR) is a challenging video retrieval task that utilizes multi-modal queries, consisting of a reference video and modification text, to retrieve the desired target video. The core of this task lies in understanding the multi-modal composed query and achieving accurate composed feature learning. Within multi-modal queries, the video modality typically carries richer semantic content compared to the textual modality. However, previous works have largely overlooked the disparity in information density between these two modalities. This limitation can lead to two critical issues: 1) modification subject referring ambiguity and 2) limited detailed semantic focus, both of which degrade the performance of CVR models. To address the aforementioned issues, we propose a novel CVR framework, namely the Hierarchical Uncertainty-aware Disambiguation network (HUD). HUD is the first framework that leverages the disparity in information density between video and text to enhance multi-modal query understanding. It comprises three key components: (a) Holistic Pronoun Disambiguation, (b) Atomistic Uncertainty Modeling, and (c) Holistic-to-Atomistic Alignment. By exploiting overlapping semantics through holistic cross-modal interaction and fine-grained semantic alignment via atomistic-level cross-modal interaction, HUD enables effective object disambiguation and enhances the focus on detailed semantics, thereby achieving precise composed feature learning. Moreover, our proposed HUD is also applicable to the Composed Image Retrieval (CIR) task and achieves state-of-the-art performance across three benchmark datasets for both CVR and CIR tasks. The codes are available on https://zivchen-ty.github.io/HUD.github.io/.
Authors: Chen Hu, Mingyu Zhou, Shuai Yuan, Zhenming Peng, Tian Pu, Xiyin Li
Abstract: Infrared small target detection (ISTD) faces two major challenges: a lack of discernible target texture and severe background clutter, which results in the background obscuring the target. To enhance targets and suppress backgrounds, we propose the Basis Decomposition Module (BDM) as an extensible and lightweight module based on basis decomposition, which decomposes a complex feature into several basis features and enhances certain information while eliminating redundancy. Extending BDM leads to a series of modules, including the Spatial Difference Decomposition Module (SD$^\mathrm{2}$M), Spatial Difference Decomposition Downsampling Module (SD$^\mathrm{3}$M), and Temporal Difference Decomposition Module (TD$^\mathrm{2}$M). Based on these modules, we develop the Spatial Difference Decomposition Network (SD$^\mathrm{2}$Net) for single-frame ISTD (SISTD) and the Spatiotemporal Difference Decomposition Network (STD$^\mathrm{2}$Net) for multi-frame ISTD (MISTD). SD$^\mathrm{2}$Net integrates SD$^\mathrm{2}$M and SD$^\mathrm{3}$M within an adapted U-shaped architecture. We employ TD$^\mathrm{2}$M to introduce motion information, which transforms SD$^\mathrm{2}$Net into STD$^\mathrm{2}$Net. Extensive experiments on SISTD and MISTD datasets demonstrate state-of-the-art (SOTA) performance. On the SISTD task, SD$^\mathrm{2}$Net performs well compared to most established networks. On the MISTD datasets, STD$^\mathrm{2}$Net achieves a mIoU of 87.68\%, outperforming SD$^\mathrm{2}$Net, which achieves a mIoU of 64.97\%. Our codes are available: https://github.com/greekinRoma/IRSTD_HC_Platform.
Authors: Jin-Ting He, Fu-Jen Tsai, Yan-Tsung Peng, Min-Hung Chen, Chia-Wen Lin, Yen-Yu Lin
Abstract: Diffusion models show promise for dynamic scene deblurring; however, existing studies often fail to leverage the intrinsic nature of the blurring process within diffusion models, limiting their full potential. To address it, we present a Blur Diffusion Model (BlurDM), which seamlessly integrates the blur formation process into diffusion for image deblurring. Observing that motion blur stems from continuous exposure, BlurDM implicitly models the blur formation process through a dual-diffusion forward scheme, diffusing both noise and blur onto a sharp image. During the reverse generation process, we derive a dual denoising and deblurring formulation, enabling BlurDM to recover the sharp image by simultaneously denoising and deblurring, given pure Gaussian noise conditioned on the blurred image as input. Additionally, to efficiently integrate BlurDM into deblurring networks, we perform BlurDM in the latent space, forming a flexible prior generation network for deblurring. Extensive experiments demonstrate that BlurDM significantly and consistently enhances existing deblurring methods on four benchmark datasets. The project page is available at https://jin-ting-he.github.io/Blur-Diffusion-Model/.
Authors: Farhana Hossain Swarnali, Miaomiao Zhang, Tonmoy Hossain
Abstract: Modeling group actions on latent representations enables controllable transformations of high-dimensional image data. Prior works applying group-theoretic priors or modeling transformations typically operate in the high-dimensional data space, where group actions apply uniformly across the entire input, making it difficult to disentangle the subspace that varies under transformations. While latent-space methods offer greater flexibility, they still require manual partitioning of latent variables into equivariant and invariant subspaces, limiting the ability to robustly learn and operate group actions within the representation space. To address this, we introduce a novel end-to-end framework that for the first time learns group actions on latent image manifolds, automatically discovering transformation-relevant structures without manual intervention. Our method uses learnable binary masks with straight-through estimation to dynamically partition latent representations into transformation-sensitive and invariant components. We formulate this within a unified optimization framework that jointly learns latent disentanglement and group transformation mappings. The framework can be seamlessly integrated with any standard encoder-decoder architecture. We validate our approach on five 2D/3D image datasets, demonstrating its ability to automatically learn disentangled latent factors for group actions in diverse data, while downstream classification tasks confirm the effectiveness of the learned representations. Our code is publicly available at https://github.com/farhanaswarnali/Learning-Group-Actions-In-Disentangled-Latent-Image-Representations .
Authors: Xin He, Longhui Wei, Jianbo Ouyang, Minghui Liao, Lingxi Xie, Qi Tian
Abstract: We propose EMMA, an efficient and unified architecture for multimodal understanding, generation and editing. Specifically, EMMA primarily consists of 1) An efficient autoencoder with a 32x compression ratio, which significantly reduces the number of tokens required for generation. This also ensures the training balance between understanding and generation tasks by applying the same compression ratio to images. 2) Channel-wise concatenation instead of token-wise concatenation among visual understanding and generation tokens, which further reduces the visual tokens in unified architectures. 3) A shared-and-decoupled network that enables mutual improvements across tasks while meeting the task-specific modeling requirements. 4) A mixture-of-experts mechanism adopted for visual understanding encoder, which substantially improves perceptual capabilities with a few parameters increase. Extensive experiments have shown that EMMA-4B can significantly outperform state-of-the-art unified multimodal approaches (e.g., BAGEL-7B) in both efficiency and performance, while also achieving competitive results compared to recent multimodal understanding and generation experts (e.g., Qwen3-VL and Qwen-Image). We believe that EMMA lays a solid foundation for the future development of unified multimodal architectures.
Authors: Tianqi Liu, Zhaoxi Chen, Zihao Huang, Shaocong Xu, Saining Zhang, Chongjie Ye, Bohan Li, Zhiguo Cao, Wei Li, Hao Zhao, Ziwei Liu
Abstract: Recent advances in illumination control extend image-based methods to video, yet still facing a trade-off between lighting fidelity and temporal consistency. Moving beyond relighting, a key step toward generative modeling of real-world scenes is the joint control of camera trajectory and illumination, since visual dynamics are inherently shaped by both geometry and lighting. To this end, we present Light-X, a video generation framework that enables controllable rendering from monocular videos with both viewpoint and illumination control. 1) We propose a disentangled design that decouples geometry and lighting signals: geometry and motion are captured via dynamic point clouds projected along user-defined camera trajectories, while illumination cues are provided by a relit frame consistently projected into the same geometry. These explicit, fine-grained cues enable effective disentanglement and guide high-quality illumination. 2) To address the lack of paired multi-view and multi-illumination videos, we introduce Light-Syn, a degradation-based pipeline with inverse-mapping that synthesizes training pairs from in-the-wild monocular footage. This strategy yields a dataset covering static, dynamic, and AI-generated scenes, ensuring robust training. Extensive experiments show that Light-X outperforms baseline methods in joint camera-illumination control and surpasses prior video relighting methods under both text- and background-conditioned settings.
Authors: Vsevolod Plohotnuk, Artyom Panshin, Nikola Bani\'c, Simone Bianco, Michael Freeman, Egor Ershov
Abstract: Traditional Image Quality Assessment~(IQA) focuses on quantifying technical degradations such as noise, blur, or compression artifacts, using both full-reference and no-reference objective metrics. However, evaluation of rendering aesthetics, a growing domain relevant to photographic editing, content creation, and AI-generated imagery, remains underexplored due to the lack of datasets that reflect the inherently subjective nature of style preference. In this work, a novel benchmark dataset designed to model human aesthetic judgments of image rendering styles is introduced: the Dataset for Evaluating the Aesthetics of Rendering (DEAR). Built upon the MIT-Adobe FiveK dataset, DEAR incorporates pairwise human preference scores collected via large-scale crowdsourcing, with each image pair evaluated by 25 distinct human evaluators with a total of 13,648 of them participating overall. These annotations capture nuanced, context-sensitive aesthetic preferences, enabling the development and evaluation of models that go beyond traditional distortion-based IQA, focusing on a new task: Evaluation of Aesthetics of Rendering (EAR). The data collection pipeline is described, human voting patterns are analyzed, and multiple use cases are outlined, including style preference prediction, aesthetic benchmarking, and personalized aesthetic modeling. To the best of the authors' knowledge, DEAR is the first dataset to systematically address image aesthetics of rendering assessment grounded in subjective human preferences. A subset of 100 images with markup for them is published on HuggingFace (huggingface.co/datasets/vsevolodpl/DEAR).
Authors: Tianyi Gao, Hao Li, Han Fang, Xin Wei, Xiaodong Dong, Hongbo Sun, Ye Yuan, Zhongjiang He, Jinglin Xu, Jingmin Xin, Hao Sun
Abstract: Referring Expression Comprehension (REC) is a vision-language task that localizes a specific image region based on a textual description. Existing REC benchmarks primarily evaluate perceptual capabilities and lack interpretable scoring mechanisms, which cannot reveal the grounding capability of Multi-modal Large Language Model (MLLM) across different cognitive abilities. To address this limitation, we introduce RefBench-PRO, a comprehensive REC benchmark, which decomposes referring expressions into two core dimensions, i.e., perception and reasoning, and further subdivides them into six progressively challenging tasks, such as attribute, position, interaction, commonsense, relation and reject. We also develop a fully automated data-generation pipeline that produces diverse referring expressions across these six sub-dimensions. Furthermore, We propose Ref-R1, an RL-based learning scheme, which incorporates Dynamic IoU-based GRPO to improve localization accuracy under increasingly complex reasoning conditions, establishing a stronger baseline for REC. Extensive experiments demonstrate that our RefBench-PRO enables interpretable evaluation of MLLM on referring expression comprehension, presenting greater challenges in both perception and reasoning.
Authors: Yutong Wang, Haiyu Zhang, Tianfan Xue, Yu Qiao, Yaohui Wang, Chang Xu, Xinyuan Chen
Abstract: The rapid development of generative models has significantly advanced image and video applications. Among these, video creation, aimed at generating videos under various conditions, has gained substantial attention. However, existing video creation models either focus solely on a few specific conditions or suffer from excessively long generation times due to complex model inference, making them impractical for real-world applications. To mitigate these issues, we propose an efficient unified video creation model, named VDOT. Concretely, we model the training process with the distribution matching distillation (DMD) paradigm. Instead of using the Kullback-Leibler (KL) minimization, we additionally employ a novel computational optimal transport (OT) technique to optimize the discrepancy between the real and fake score distributions. The OT distance inherently imposes geometric constraints, mitigating potential zero-forcing or gradient collapse issues that may arise during KL-based distillation within the few-step generation scenario, and thus, enhances the efficiency and stability of the distillation process. Further, we integrate a discriminator to enable the model to perceive real video data, thereby enhancing the quality of generated videos. To support training unified video creation models, we propose a fully automated pipeline for video data annotation and filtering that accommodates multiple video creation tasks. Meanwhile, we curate a unified testing benchmark, UVCBench, to standardize evaluation. Experiments demonstrate that our 4-step VDOT outperforms or matches other baselines with 100 denoising steps.
Authors: Dahyeon Kye, Jeahun Sung, Mingyu Jeon, Jihyong Oh
Abstract: Diffusion models exhibit remarkable generative ability, yet achieving smooth and semantically consistent image morphing remains a challenge. Existing approaches often yield abrupt transitions or over-saturated appearances due to the lack of adaptive structural and semantic alignments. We propose CHIMERA, a zero-shot diffusion-based framework that formulates morphing as a cached inversion-guided denoising process. To handle large semantic and appearance disparities, we propose Adaptive Cache Injection and Semantic Anchor Prompting. Adaptive Cache Injection (ACI) caches down, mid, and up blocks features from both inputs during DDIM inversion and re-injects them adaptively during denoising, enabling spatial and semantic alignment in depth- and time-adaptive manners and enabling natural feature fusion and smooth transitions. Semantic Anchor Prompting (SAP) leverages a vision-language model to generate a shared anchor prompt that serves as a semantic anchor, bridging dissimilar inputs and guiding the denoising process toward coherent results. Finally, we introduce the Global-Local Consistency Score (GLCS), a morphing-oriented metric that simultaneously evaluates the global harmonization of the two inputs and the smoothness of the local morphing transition. Extensive experiments and user studies show that CHIMERA achieves smoother and more semantically aligned transitions than existing methods, establishing a new state of the art in image morphing. The code and project page will be publicly released.
Authors: Chang Liu, Hongliang Yuan, Lianghao Zhang, Sichao Wang, Jianwei Guo, Shi-Sheng Huang
Abstract: Rendering complex reflection of real-world scenes using 3D Gaussian splatting has been a quite promising solution for photorealistic novel view synthesis, but still faces bottlenecks especially in rendering speed and memory storage. This paper proposes a new Hybrid Splatting(HybridSplat) mechanism for Gaussian primitives. Our key idea is a new reflection-baked Gaussian tracing, which bakes the view-dependent reflection within each Gaussian primitive while rendering the reflection using tile-based Gaussian splatting. Then we integrate the reflective Gaussian primitives with base Gaussian primitives using a unified hybrid splatting framework for high-fidelity scene reconstruction. Moreover, we further introduce a pipeline-level acceleration for the hybrid splatting, and reflection-sensitive Gaussian pruning to reduce the model size, thus achieving much faster rendering speed and lower memory storage while preserving the reflection rendering quality. By extensive evaluation, our HybridSplat accelerates about 7x rendering speed across complex reflective scenes from Ref-NeRF, NeRF-Casting with 4x fewer Gaussian primitives than similar ray-tracing based Gaussian splatting baselines, serving as a new state-of-the-art method especially for complex reflective scenes.
Authors: Shaofeng Zhang, Xuanqi Chen, Ning Liao, Haoxiang Zhao, Xiaoxing Wang, Haoru Tan, Sitong Wu, Xiaosong Jia, Qi Fan, Junchi Yan
Abstract: The dominance of denoising generative models (e.g., diffusion, flow-matching) in visual synthesis is tempered by their substantial training costs and inefficiencies in representation learning. While injecting discriminative representations via auxiliary alignment has proven effective, this approach still faces key limitations: the reliance on external, pre-trained encoders introduces overhead and domain shift. A dispersed-based strategy that encourages strong separation among in-batch latent representations alleviates this specific dependency. To assess the effect of the number of negative samples in generative modeling, we propose {\mname}, a plug-and-play training framework that requires no external encoders. Our method integrates a memory bank mechanism that maintains a large, dynamically updated queue of negative samples across training iterations. This decouples the number of negatives from the mini-batch size, providing abundant and high-quality negatives for a contrastive objective without a multiplicative increase in computational cost. A low-dimensional projection head is used to further minimize memory and bandwidth overhead. {\mname} offers three principal advantages: (1) it is self-contained, eliminating dependency on pretrained vision foundation models and their associated forward-pass overhead; (2) it introduces no additional parameters or computational cost during inference; and (3) it enables substantially faster convergence, achieving superior generative quality more efficiently. On ImageNet-256, {\mname} achieves a state-of-the-art FID of \textbf{2.40} within 400k steps, significantly outperforming comparable methods.
Authors: Manduhu Manduhu, Alexander Dow, Gerard Dooly, James Riordan
Abstract: Rotation invariance is essential for precise, object-level segmentation in UAV aerial imagery, where targets can have arbitrary orientations and exhibit fine-scale details. Conventional segmentation architectures like U-Net rely on convolution operators that are not rotation-invariant, leading to degraded segmentation accuracy across varying viewpoints. Rotation invariance can be achieved by expanding the filter bank across multiple orientations; however, this will significantly increase computational cost and memory traffic. In this paper, we introduce a GPU-optimized rotation-invariant convolution framework that eliminates the traditional data-lowering (im2col) step required for matrix-multiplication-based convolution. By exploiting structured data sharing among symmetrically rotated filters, our method achieves multi-orientation convolution with greatly reduced memory traffic and computational redundancy. We further generalize the approach to accelerate convolution with arbitrary (non-symmetric) rotation angles. Across extensive benchmarks, the proposed convolution achieves 20--55% faster training and 15--45% lower energy consumption than CUDNN, while maintaining accuracy comparable to state-of-the-art rotation-invariant methods. In the eight-orientation setting, our approach achieves up to 45% speedup and 41% energy savings on 256\(\times\)256 inputs, and 32% speedup and 23% lower energy usage on 1024\(\times\)1024 inputs. Integrated into a U-Net segmentation model, the framework yields up to 6% improvement in accuracy over the non-rotation-aware baseline. These results demonstrate that the proposed method provides an effective and highly efficient alternative to existing rotation-invariant CNN frameworks.
Authors: Yixuan Zhu, Jiaqi Feng, Wenzhao Zheng, Yuan Gao, Xin Tao, Pengfei Wan, Jie Zhou, Jiwen Lu
Abstract: Recent advances in diffusion transformers have empowered video generation models to generate high-quality video clips from texts or images. However, world models with the ability to predict long-horizon futures from past observations and actions remain underexplored, especially for general-purpose scenarios and various forms of actions. To bridge this gap, we introduce Astra, an interactive general world model that generates real-world futures for diverse scenarios (e.g., autonomous driving, robot grasping) with precise action interactions (e.g., camera motion, robot action). We propose an autoregressive denoising architecture and use temporal causal attention to aggregate past observations and support streaming outputs. We use a noise-augmented history memory to avoid over-reliance on past frames to balance responsiveness with temporal coherence. For precise action control, we introduce an action-aware adapter that directly injects action signals into the denoising process. We further develop a mixture of action experts that dynamically route heterogeneous action modalities, enhancing versatility across diverse real-world tasks such as exploration, manipulation, and camera control. Astra achieves interactive, consistent, and general long-term video prediction and supports various forms of interactions. Experiments across multiple datasets demonstrate the improvements of Astra in fidelity, long-range prediction, and action alignment over existing state-of-the-art world models.
Authors: Hao Chen, Rui Yin, Yifan Chen, Qi Chen, Chao Li
Abstract: Understanding disease progression is a central clinical challenge with direct implications for early diagnosis and personalized treatment. While recent generative approaches have attempted to model progression, key mismatches remain: disease dynamics are inherently continuous and monotonic, yet latent representations are often scattered, lacking semantic structure, and diffusion-based models disrupt continuity with random denoising process. In this work, we propose to treat the disease dynamic as a velocity field and leverage Flow Matching (FM) to align the temporal evolution of patient data. Unlike prior methods, it captures the intrinsic dynamic of disease, making the progression more interpretable. However, a key challenge remains: in latent space, Auto-Encoders (AEs) do not guarantee alignment across patients or correlation with clinical-severity indicators (e.g., age and disease conditions). To address this, we propose to learn patient-specific latent alignment, which enforces patient trajectories to lie along a specific axis, with magnitude increasing monotonically with disease severity. This leads to a consistent and semantically meaningful latent space. Together, we present $\Delta$-LFM, a framework for modeling patient-specific latent progression with flow matching. Across three longitudinal MRI benchmarks, $\Delta$-LFM demonstrates strong empirical performance and, more importantly, offers a new framework for interpreting and visualizing disease dynamics.
Authors: Kanghyun Baek, Sangyub Lee, Jin Young Choi, Jaewoo Song, Daemin Park, Jooyoung Choi, Chaehun Shin, Bohyung Han, Sungroh Yoon
Abstract: Despite recent advances, diffusion-based text-to-image models still struggle with accurate text rendering. Several studies have proposed fine-tuning or training-free refinement methods for accurate text rendering. However, the critical issue of text omission, where the desired text is partially or entirely missing, remains largely overlooked. In this work, we propose TextGuider, a novel training-free method that encourages accurate and complete text appearance by aligning textual content tokens and text regions in the image. Specifically, we analyze attention patterns in Multi-Modal Diffusion Transformer(MM-DiT) models, particularly for text-related tokens intended to be rendered in the image. Leveraging this observation, we apply latent guidance during the early stage of denoising steps based on two loss functions that we introduce. Our method achieves state-of-the-art performance in test-time text rendering, with significant gains in recall and strong results in OCR accuracy and CLIP score.
Authors: Zhe Li, Hadrien Reynaud, Alberto Gomez, Bernhard Kainz
Abstract: Echocardiography plays a critical role in the diagnosis and monitoring of cardiovascular diseases as a non-invasive real-time assessment of cardiac structure and function. However, the growing scale of echocardiographic video data presents significant challenges in terms of storage, computation, and model training efficiency. Dataset distillation offers a promising solution by synthesizing a compact, informative subset of data that retains the key clinical features of the original dataset. In this work, we propose a novel approach for distilling a compact synthetic echocardiographic video dataset. Our method leverages motion feature extraction to capture temporal dynamics, followed by class-wise graph construction and representative sample selection using the Infomap algorithm. This enables us to select a diverse and informative subset of synthetic videos that preserves the essential characteristics of the original dataset. We evaluate our approach on the EchoNet-Dynamic datasets and achieve a test accuracy of \(69.38\%\) using only \(25\) synthetic videos. These results demonstrate the effectiveness and scalability of our method for medical video dataset distillation.
Authors: Yixin Wan, Lei Ke, Wenhao Yu, Kai-Wei Chang, Dong Yu
Abstract: We introduce MotionEdit, a novel dataset for motion-centric image editing-the task of modifying subject actions and interactions while preserving identity, structure, and physical plausibility. Unlike existing image editing datasets that focus on static appearance changes or contain only sparse, low-quality motion edits, MotionEdit provides high-fidelity image pairs depicting realistic motion transformations extracted and verified from continuous videos. This new task is not only scientifically challenging but also practically significant, powering downstream applications such as frame-controlled video synthesis and animation. To evaluate model performance on the novel task, we introduce MotionEdit-Bench, a benchmark that challenges models on motion-centric edits and measures model performance with generative, discriminative, and preference-based metrics. Benchmark results reveal that motion editing remains highly challenging for existing state-of-the-art diffusion-based editing models. To address this gap, we propose MotionNFT (Motion-guided Negative-aware Fine Tuning), a post-training framework that computes motion alignment rewards based on how well the motion flow between input and model-edited images matches the ground-truth motion, guiding models toward accurate motion transformations. Extensive experiments on FLUX.1 Kontext and Qwen-Image-Edit show that MotionNFT consistently improves editing quality and motion fidelity of both base models on the motion editing task without sacrificing general editing ability, demonstrating its effectiveness. Our code is at https://github.com/elainew728/motion-edit/.
Authors: Jiawei Yang, Ziyu Chen, Yurong You, Yan Wang, Yiming Li, Yuxiao Chen, Boyi Li, Boris Ivanovic, Marco Pavone, Yue Wang
Abstract: We present Flex, an efficient and effective scene encoder that addresses the computational bottleneck of processing high-volume multi-camera data in end-to-end autonomous driving. Flex employs a small set of learnable scene tokens to jointly encode information from all image tokens across different cameras and timesteps. By design, our approach is geometry-agnostic, learning a compact scene representation directly from data without relying on the explicit 3D inductive biases, such as Bird-Eye-View (BEV), occupancy or tri-plane representations, which are common in prior work. This holistic encoding strategy aggressively compresses the visual input for the downstream Large Language Model (LLM) based policy model. Evaluated on a large-scale proprietary dataset of 20,000 driving hours, our Flex achieves 2.2x greater inference throughput while improving driving performance by a large margin compared to state-of-the-art methods. Furthermore, we show that these compact scene tokens develop an emergent capability for scene decomposition without any explicit supervision. Our findings challenge the prevailing assumption that 3D priors are necessary, demonstrating that a data-driven, joint encoding strategy offers a more scalable, efficient and effective path for future autonomous driving systems.
Authors: Zhengyang Yu, Akio Hayakawa, Masato Ishii, Qingtao Yu, Takashi Shibuya, Jing Zhang, Yuki Mitsufuji
Abstract: Autoregressive video diffusion models (AR-VDMs) show strong promise as scalable alternatives to bidirectional VDMs, enabling real-time and interactive applications. Yet there remains room for improvement in their sample fidelity. A promising solution is inference-time alignment, which optimizes the noise space to improve sample fidelity without updating model parameters. Yet, optimization- or search-based methods are computationally impractical for AR-VDMs. Recent text-to-image (T2I) works address this via feedforward noise refiners that modulate sampled noises in a single forward pass. Can such noise refiners be extended to AR-VDMs? We identify the failure of naively extending T2I noise refiners to AR-VDMs and propose AutoRefiner-a noise refiner tailored for AR-VDMs, with two key designs: pathwise noise refinement and a reflective KV-cache. Experiments demonstrate that AutoRefiner serves as an efficient plug-in for AR-VDMs, effectively enhancing sample fidelity by refining noise along stochastic denoising paths.
Authors: Fei Long, Yao Zhang, Jiaming Lv, Jiangtao Xie, Peihua Li
Abstract: Few-shot action recognition (FSAR) has recently made notable progress through set matching and efficient adaptation of large-scale pre-trained models. However, two key limitations persist. First, existing set matching metrics typically rely on cosine similarity to measure inter-frame linear dependencies and then perform matching with only instance-level information, thus failing to capture more complex patterns such as nonlinear relationships and overlooking task-specific cues. Second, for efficient adaptation of CLIP to FSAR, recent work performing fine-tuning via skip-fusion layers (which we refer to as side layers) has significantly reduced memory cost. However, the newly introduced side layers are often difficult to optimize under limited data conditions. To address these limitations, we propose TS-FSAR, a framework comprising three components: (1) a visual Ladder Side Network (LSN) for efficient CLIP fine-tuning; (2) a metric called Task-Specific Distance Correlation Matching (TS-DCM), which uses $\alpha$-distance correlation to model both linear and nonlinear inter-frame dependencies and leverages a task prototype to enable task-specific matching; and (3) a Guiding LSN with Adapted CLIP (GLAC) module, which regularizes LSN using the adapted frozen CLIP to improve training for better $\alpha$-distance correlation estimation under limited supervision. Extensive experiments on five widely-used benchmarks demonstrate that our TS-FSAR yields superior performance compared to prior state-of-the-arts.
Authors: Ahmed Mujtaba, Gleb Radchenko, Marc Masana, Radu Prodan
Abstract: Driver fatigue remains a leading cause of road accidents, with 24% of crashes involving drowsy drivers. While yawning serves as an early behavioral indicator of fatigue, existing machine learning approaches face significant challenges due to video-annotated datasets that introduce systematic noise from coarse temporal annotations. We develop a semi-automated labeling pipeline with human-in-the-loop verification, which we apply to YawDD, enabling more accurate model training. Training the established MNasNet classifier and YOLOv11 detector architectures on YawDD+ improves frame accuracy by up to 6% and mAP by 5% over video-level supervision, achieving 99.34% classification accuracy and 95.69% detection mAP. The resulting approach deliver up to 59.8 FPS on edge AI hardware (NVIDIA Jetson Nano), confirming that enhanced data quality alone supports on-device yawning monitoring without server-side computation.
Authors: Lujia Zhong, Shuo Huang, Jiaxin Yue, Jianwei Zhang, Zhiwei Deng, Wenhao Chi, Yonggang Shi
Abstract: The emergence of tau PET imaging over the last decade has enabled Alzheimer's disease (AD) researchers to examine tau pathology in vivo and more effectively characterize the disease trajectories of AD. Current tau PET analysis methods, however, typically perform inferences on large cortical ROIs and are limited in the detection of localized tau pathology that varies across subjects. In this work, we propose a novel bilateral-guided deterministic diffusion sampling method to perform anomaly detection from tau PET imaging data. By including individualized brain structure and cognitively normal (CN) template conditions, our model computes a voxel-level anomaly map based on the deterministically sampled pseudo-healthy reconstruction. We train our model on ADNI CN subjects (n=380) and evaluate anomaly localization performance on the left MCI/AD subjects (n=154) and the preclinical subjects of the A4 clinical trial (n=447). We further train a CNN classifier on the derived 3D anomaly maps from ADNI, including CN and MCI/AD, to classify subjects into two groups and test classification performance on A4. We demonstrate that our method outperforms baselines in anomaly localization. Additionally, we show that our method can successfully group preclinical subjects with significantly different cognitive functions, highlighting the potential of our approach for application in preclinical screening tests. The code will be publicly available.
Authors: Honggui Li, Sinan Chen, Nahid Md Lokman Hossain, Maria Trocan, Dimitri Galayko, Mohamad Sawan
Abstract: Learned image compression (LIC) is currently the cutting-edge method. However, the inherent difference between testing and training images of LIC results in performance degradation to some extent. Especially for out-of-sample, out-of-distribution, or out-of-domain testing images, the performance of LIC dramatically degraded. Classical LIC is a serial image compression (SIC) approach that utilizes an open-loop architecture with serial encoding and decoding units. Nevertheless, according to the theory of automatic control, a closed-loop architecture holds the potential to improve the dynamic and static performance of LIC. Therefore, a circular image compression (CIC) approach with closed-loop encoding and decoding elements is proposed to minimize the gap between testing and training images and upgrade the capability of LIC. The proposed CIC establishes a nonlinear loop equation and proves that steady-state error between reconstructed and original images is close to zero by Taylor series expansion. The proposed CIC method possesses the property of Post-Training and plug-and-play which can be built on any existing advanced SIC methods. Experimental results on five public image compression datasets demonstrate that the proposed CIC outperforms five competing state-of-the-art open-source SIC algorithms in reconstruction capacity. Experimental results further show that the proposed method is suitable for out-of-sample testing images with dark backgrounds, sharp edges, high contrast, grid shapes, or complex patterns.
Authors: Shoujin Huang, Guanxiong Luo, Yunlin Zhao, Yilong Liu, Yuwan Wang, Kexin Yang, Jingzhe Liu, Hua Guo, Min Wang, Lingyan Zhang, Mengye Lyu
Abstract: Simultaneous multislice (SMS) imaging is a powerful technique for accelerating magnetic resonance imaging (MRI) acquisitions. However, SMS reconstruction remains challenging due to complex signal interactions between and within the excited slices. In this study, we introduce ROGER, a robust SMS MRI reconstruction method based on deep generative priors. Utilizing denoising diffusion probabilistic models (DDPM), ROGER begins with Gaussian noise and gradually recovers individual slices through reverse diffusion iterations while enforcing data consistency from measured k-space data within the readout concatenation framework. The posterior sampling procedure is designed such that the DDPM training can be performed on single-slice images without requiring modifications for SMS tasks. Additionally, our method incorporates a low-frequency enhancement (LFE) module to address the practical issue that SMS-accelerated fast spin echo (FSE) and echo planar imaging (EPI) sequences cannot easily embed fully-sampled autocalibration signals. Extensive experiments on both retrospectively and prospectively accelerated datasets demonstrate that ROGER consistently outperforms existing methods, enhancing both anatomical and functional imaging with strong out-of-distribution generalization. The source code and sample data for ROGER are available at https://github.com/Solor-pikachu/ROGER.
Authors: Pengfei Guo, Can Zhao, Dong Yang, Ziyue Xu, Vishwesh Nath, Yucheng Tang, Benjamin Simon, Mason Belue, Stephanie Harmon, Baris Turkbey, Daguang Xu
Abstract: Medical imaging analysis faces challenges such as data scarcity, high annotation costs, and privacy concerns. This paper introduces the Medical AI for Synthetic Imaging (MAISI), an innovative approach using the diffusion model to generate synthetic 3D computed tomography (CT) images to address those challenges. MAISI leverages the foundation volume compression network and the latent diffusion model to produce high-resolution CT images (up to a landmark volume dimension of 512 x 512 x 768 ) with flexible volume dimensions and voxel spacing. By incorporating ControlNet, MAISI can process organ segmentation, including 127 anatomical structures, as additional conditions and enables the generation of accurately annotated synthetic images that can be used for various downstream tasks. Our experiment results show that MAISI's capabilities in generating realistic, anatomically accurate images for diverse regions and conditions reveal its promising potential to mitigate challenges using synthetic data.
Authors: Paul Weiser, Georg Langs, Stanislav Motyka, Wolfgang Bogner, S\'ebastien Courvoisier, Malte Hoffmann, Antoine Klauser, Ovidiu C. Andronesi
Abstract: Purpose. Proton Magnetic Resonance Spectroscopic Imaging (1H-MRSI) provides non-invasive spectral-spatial mapping of metabolism. However, long-standing problems in whole-brain 1H-MRSI are spectral overlap of metabolite peaks with large lipid signal from scalp, and overwhelming water signal that distorts spectra. Fast and effective methods are needed for high-resolution 1H-MRSI to accurately remove lipid and water signals while preserving the metabolite signal. The potential of supervised neural networks for this task remains unexplored, despite their success for other MRSI processing. Methods. We introduce a deep-learning method based on a modified Y-NET network for water and lipid removal in whole-brain 1H-MRSI. The WALINET (WAter and LIpid neural NETwork) was compared to conventional methods such as the state-of-the-art lipid L2 regularization and Hankel-Lanczos singular value decomposition (HLSVD) water suppression. Methods were evaluated on simulated and in-vivo whole-brain MRSI using NMRSE, SNR, CRLB, and FWHM metrics. Results. WALINET is significantly faster and needs 8s for high-resolution whole-brain MRSI, compared to 42 minutes for conventional HLSVD+L2. Quantitative analysis shows WALINET has better performance than HLSVD+L2: 1) more lipid removal with 41% lower NRMSE, 2) better metabolite signal preservation with 71% lower NRMSE in simulated data, 155% higher SNR and 50% lower CRLB in in-vivo data. Metabolic maps obtained by WALINET in healthy subjects and patients show better gray/white-matter contrast with more visible structural details. Conclusions. WALINET has superior performance for nuisance signal removal and metabolite quantification on whole-brain 1H-MRSI compared to conventional state-of-the-art techniques. This represents a new application of deep-learning for MRSI processing, with potential for automated high-throughput workflow.
Authors: Tomasz Rybotycki, Manish K. Gupta, Piotr Gawron
Abstract: The emergence of Big Data changed how we approach information systems engineering. Nowadays, when we can use remote sensing techniques for Big Data acquisition, the issues such data introduce are as important as ever. One of those concerns is the processing of the data. Classical methods often fail to address that problem or are incapable of processing the data in a reasonable time. With that in mind information system engineers are required to investigate different approaches to the data processing. The recent advancements in noisy intermediate-scale quantum (NISQ) devices implementation allow us to investigate their application to real-life computational problem. This field of study is called quantum (information) systems engineering and usually focuses on technical problems with the contemporary devices. However, hardware challenges are not the only ones that hinder our quantum computation capabilities. Software limitations are the other, less explored side of this medal. Using multispectral image segmentation as a task example, we investigated how difficult it is to run a hybrid quantum-classical model on a real, publicly available quantum device. To quantify how and explain why the performance of our model changed when ran on a real device, we propose new explainability metrics. These metrics introduce new meaning to the explainable quantum machine learning; the explanation of the performance issue comes from the quantum device behavior. We also analyzed the expected money costs of running similar experiment on contemporary quantum devices using standard market prices.
Authors: Lin Tian, Jonathan Williams-Ramirez, Dina Zemlyanker, Lucas J. Deden-Binder, Rogeny Herisse, Theresa R. Connors, Mark Montine, Istvan N Huszar, Lilla Z\"ollei, Sean I. Young, Christine Mac Donald, C. Dirk Keene, Derek H. Oakley, Bradley T. Hyman, Oula Puonti, Matthew S. Rosen, Juan Eugenio Iglesias
Abstract: Correlation of neuropathology with MRI has the potential to transfer microscopic signatures of pathology to in vivo scans. There is increasing interest in building these correlations from 3D reconstructed stacks of slab photographs, which are routinely taken during dissection at brain banks. These photographs bypass the need for ex vivo MRI, which is not widely accessible. However, existing methods either require a corresponding 3D reference (e.g., an ex vivo MRI scans, or a brain surface acquired with a structured light scanner) or a full stack of brain slabs, which severely limits applicability. Here we propose RefFree, a 3D reconstruction method for dissection photographs that does not require an external reference. RefFree coherently reconstructs a 3D volume for an arbitrary set of slabs (including a single slab) using predicted 3D coordinates in the standard atlas space (MNI) as guidance. To support RefFree's pipeline, we train an atlas coordinate prediction network that estimates the coordinate map from a 2D photograph, using synthetic photographs generated from digitally sliced 3D MRI data with randomized appearance for enhanced generalization. As a by-product, RefFree can propagate information (e.g., anatomical labels) from atlas space to one single photograph even without reconstruction. Experiments on simulated and real data show that, when all slabs are available, RefFree achieves performance comparable to existing classical methods but at substantially higher speed. Moreover, RefFree yields accurate reconstruction and registration for partial stacks or even a single slab. Our code is available at https://github.com/lintian-a/reffree.
Authors: Kangwei Liu, Mengru Wang, Yujie Luo, Lin Yuan, Mengshu Sun, Lei Liang, Zhiqiang Zhang, Jun Zhou, Bryan Hooi, Shumin Deng
Abstract: Fine-tuning enables large language models (LLMs) to adapt to specific domains, but often compromises their previously established safety alignment. To mitigate the degradation of model safety during fine-tuning, we introduce LookAhead Tuning, a lightweight and effective data-driven approach that preserves safety during fine-tuning. The method introduces two simple strategies that modify training data by previewing partial answer prefixes, thereby minimizing perturbations to the model's initial token distributions and maintaining its built-in safety mechanisms. Comprehensive experiments demonstrate that LookAhead Tuning effectively maintains model safety without sacrificing robust performance on downstream tasks. Our findings position LookAhead Tuning as a reliable and efficient solution for the safe and effective adaptation of LLMs.
Authors: Chang Cai, Xiaojun Yuan, Ying-Jun Angela Zhang
Abstract: Message passing algorithms have been tailored for compressive imaging applications by plugging in different types of off-the-shelf image denoisers. These off-the-shelf denoisers mostly rely on some generic or hand-crafted priors for denoising. Due to their insufficient accuracy in capturing the true image prior, these methods often fail to produce satisfactory results, especially in highly underdetermined scenarios. On the other hand, score-based generative modeling offers a promising way to accurately characterize the sophisticated image distribution. In this paper, by exploiting the close relation between score-based modeling and empirical Bayes-optimal denoising, we devise a message passing framework that integrates a score-based minimum mean squared error (MMSE) denoiser for compressive image recovery. Experiments on the FFHQ dataset demonstrate that our method strikes a significantly better performance-complexity tradeoff than conventional message passing, regularized linear regression, and score-based posterior sampling baselines. Remarkably, our method typically converges in fewer than 20 neural function evaluations (NFEs).
Authors: Kazuki Higo, Toshiki Kanai, Yuki Endo, Yoshihiro Kanamori
Abstract: 3D terrain models are essential in fields such as video game development and film production. Since surface color often correlates with terrain geometry, capturing this relationship is crucial to achieving realism. However, most existing methods generate either a heightmap or a texture, without sufficiently accounting for the inherent correlation. In this paper, we propose a method that jointly generates terrain heightmaps and textures using a latent diffusion model. First, we train the model in an unsupervised manner to randomly generate paired heightmaps and textures. Then, we perform supervised learning of an external adapter to enable user control via hand-drawn sketches. Experiments show that our approach allows intuitive terrain generation while preserving the correlation between heightmaps and textures.
Authors: Kangxian Xie, Yufei Zhu, Kaiming Kuang, Li Zhang, Hongwei Bran Li, Mingchen Gao, Jiancheng Yang
Abstract: High-quality 3D reconstruction of pulmonary segments plays a crucial role in segmentectomy and surgical planning for the treatment of lung cancer. Due to the resolution requirement of the target reconstruction, conventional deep learning-based methods often suffer from computational resource constraints or limited granularity. Conversely, implicit modeling is favored due to its computational efficiency and continuous representation at any resolution. We propose a neural implicit function-based method to learn a 3D surface to achieve anatomy-aware, precise pulmonary segment reconstruction, represented as a shape by deforming a learnable template. Additionally, we introduce two clinically relevant evaluation metrics to comprehensively assess the quality of the reconstruction. Furthermore, to address the lack of publicly available shape datasets for benchmarking reconstruction algorithms, we developed a shape dataset named Lung3D, which includes the 3D models of 800 labeled pulmonary segments and their corresponding airways, arteries, veins, and intersegmental veins. We demonstrate that the proposed approach outperforms existing methods, providing a new perspective for pulmonary segment reconstruction. Code and data will be available at https://github.com/HINTLab/ImPulSe.
Authors: Jiajun Jiang, Yiming Zhu, Zirui Wu, Jie Song
Abstract: We introduce DualMap, an online open-vocabulary mapping system that enables robots to understand and navigate dynamically changing environments through natural language queries. Designed for efficient semantic mapping and adaptability to changing environments, DualMap meets the essential requirements for real-world robot navigation applications. Our proposed hybrid segmentation frontend and object-level status check eliminate the costly 3D object merging required by prior methods, enabling efficient online scene mapping. The dual-map representation combines a global abstract map for high-level candidate selection with a local concrete map for precise goal-reaching, effectively managing and updating dynamic changes in the environment. Through extensive experiments in both simulation and real-world scenarios, we demonstrate state-of-the-art performance in 3D open-vocabulary segmentation, efficient scene mapping, and online language-guided navigation. Project page: https://eku127.github.io/DualMap/
Authors: Yewei Liu, Xiyuan Wang, Muhan Zhang
Abstract: We propose an entirely new meta-learning framework for network pruning. It is a general framework that can be theoretically applied to almost all types of networks with all kinds of pruning and has great generality and transferability. Experiments have shown that it can achieve outstanding results on many popular and representative pruning tasks (including both CNNs and Transformers). Unlike all prior works that either rely on fixed, hand-crafted criteria to prune in a coarse manner, or employ learning to prune ways that require special training during each pruning and lack generality. Our framework can learn complex pruning rules automatically via a neural network (metanetwork) and has great generality that can prune without any special training. More specifically, we introduce the newly developed idea of metanetwork from meta-learning into pruning. A metanetwork is a network that takes another network as input and produces a modified network as output. In this paper, we first establish a bijective mapping between neural networks and graphs, and then employ a graph neural network as our metanetwork. We train a metanetwork that learns the pruning strategy automatically and can transform a network that is hard to prune into another network that is much easier to prune. Once the metanetwork is trained, our pruning needs nothing more than a feedforward through the metanetwork and some standard finetuning to prune at state-of-the-art. Our code is available at https://github.com/Yewei-Liu/MetaPruning
Authors: Xianzhe Fan, Xuhui Zhou, Chuanyang Jin, Kolby Nottingham, Hao Zhu, Maarten Sap
Abstract: Humans continuously infer the states, goals, and behaviors of others by perceiving their surroundings in dynamic, real-world social interactions. However, most Theory of Mind (ToM) benchmarks only evaluate static, text-based scenarios, which have a significant gap compared to real interactions. We propose the SoMi-ToM benchmark, designed to evaluate multi-perspective ToM in embodied multi-agent complex social interactions. This benchmark is based on rich multimodal interaction data generated by the interaction environment SoMi, covering diverse crafting goals and social relationships. Our framework supports multi-level evaluation: (1) first-person evaluation provides multimodal (visual, dialogue, action, etc.) input from a first-person perspective during a task for real-time state inference, (2) third-person evaluation provides complete third-person perspective video and text records after a task for goal and behavior inference. This evaluation method allows for a more comprehensive examination of a model's ToM capabilities from both the subjective immediate experience and the objective global observation. We constructed a challenging dataset containing 35 third-person perspective videos, 363 first-person perspective images, and 1225 expert-annotated multiple-choice questions (three options). On this dataset, we systematically evaluated the performance of human subjects and several state-of-the-art large vision-language models (LVLMs). The results show that LVLMs perform significantly worse than humans on SoMi-ToM: the average accuracy gap between humans and models is 40.1% in first-person evaluation and 26.4% in third-person evaluation. This indicates that future LVLMs need to further improve their ToM capabilities in embodied, complex social interactions.
Authors: Yinuo Wang, Gavin Tao
Abstract: We introduce LocoMamba, a vision-driven cross-modal DRL framework built on selective state-space models, specifically leveraging Mamba, that achieves near-linear-time sequence modeling, effectively captures long-range dependencies, and enables efficient training with longer sequences. First, we embed proprioceptive states with a multilayer perceptron and patchify depth images with a lightweight convolutional neural network, producing compact tokens that improve state representation. Second, stacked Mamba layers fuse these tokens via near-linear-time selective scanning, reducing latency and memory footprint, remaining robust to token length and image resolution, and providing an inductive bias that mitigates overfitting. Third, we train the policy end-to-end with Proximal Policy Optimization under terrain and appearance randomization and an obstacle-density curriculum, using a compact state-centric reward that balances progress, smoothness, and safety. We evaluate our method in challenging simulated environments with static and moving obstacles as well as uneven terrain. Compared with state-of-the-art baselines, our method achieves higher returns and success rates with fewer collisions, exhibits stronger generalization to unseen terrains and obstacle densities, and improves training efficiency by converging in fewer updates under the same compute budget.
Authors: Hyun-Jic Oh, Junsik Kim, Zhiyi Shi, Yichen Wu, Yu-An Chen, Peter K Sorger, Hanspeter Pfister, Won-Ki Jeong
Abstract: Multiplex imaging is revolutionizing pathology by enabling the simultaneous visualization of multiple biomarkers within tissue samples, providing molecular-level insights that traditional hematoxylin and eosin (H&E) staining cannot provide. However, the complexity and cost of multiplex data acquisition have hindered its widespread adoption. Additionally, most existing large repositories of H&E images lack corresponding multiplex images, limiting opportunities for multimodal analysis. To address these challenges, we leverage recent advances in latent diffusion models (LDMs), which excel at modeling complex data distributions by utilizing their powerful priors for fine-tuning to a target domain. In this paper, we introduce a novel framework for virtual multiplex staining that utilizes pretrained LDM parameters to generate multiplex images from H&E images using a conditional diffusion model. Our approach enables marker-by-marker generation by conditioning the diffusion model on each marker, while sharing the same architecture across all markers. To tackle the challenge of varying pixel value distributions across different marker stains and to improve inference speed, we fine-tune the model for single-step sampling, enhancing both color contrast fidelity and inference efficiency through pixel-level loss functions. We validate our framework on two publicly available datasets, notably demonstrating its effectiveness in generating up to 18 different marker types with improved accuracy, a substantial increase over the 2-3 marker types achieved in previous approaches. This validation highlights the potential of our framework, pioneering virtual multiplex staining. Finally, this paper bridges the gap between H&E and multiplex imaging, potentially enabling retrospective studies and large-scale analyses of existing H&E image repositories.
Authors: Rohan Sanda, Asad Aali, Andrew Johnston, Eduardo Reis, Gordon Wetzstein, Sara Fridovich-Keil
Abstract: Magnetic resonance imaging (MRI) requires long acquisition times, raising costs, reducing accessibility, and making scans more susceptible to motion artifacts. Diffusion probabilistic models that learn data-driven priors can potentially assist in reducing acquisition time. However, they typically require large training datasets that can be prohibitively expensive to collect. Patch-based diffusion models have shown promise in learning effective data-driven priors over small real-valued datasets, but have not yet demonstrated clinical value in MRI. We extend the Patch-based Diffusion Inverse Solver (PaDIS) to complex-valued, multi-coil MRI reconstruction, and compare it against a state-of-the-art whole-image diffusion baseline (FastMRI-EDM) for 7x undersampled MRI reconstruction on the FastMRI brain dataset. We show that PaDIS-MRI models trained on small datasets of as few as 25 k-space images outperform FastMRI-EDM on image quality metrics (PSNR, SSIM, NRMSE), pixel-level uncertainty, cross-contrast generalization, and robustness to severe k-space undersampling. In a blinded study with three radiologists, PaDIS-MRI reconstructions were chosen as diagnostically superior in 91.7% of cases, compared to baselines (i) FastMRI-EDM and (ii) classical convex reconstruction with wavelet sparsity. These findings highlight the potential of patch-based diffusion priors for high-fidelity MRI reconstruction in data-scarce clinical settings where diagnostic confidence matters.
Authors: Hansheng Chen, Kai Zhang, Hao Tan, Leonidas Guibas, Gordon Wetzstein, Sai Bi
Abstract: Few-step diffusion or flow-based generative models typically distill a velocity-predicting teacher into a student that predicts a shortcut towards denoised data. This format mismatch has led to complex distillation procedures that often suffer from a quality-diversity trade-off. To address this, we propose policy-based flow models ($\pi$-Flow). $\pi$-Flow modifies the output layer of a student flow model to predict a network-free policy at one timestep. The policy then produces dynamic flow velocities at future substeps with negligible overhead, enabling fast and accurate ODE integration on these substeps without extra network evaluations. To match the policy's ODE trajectory to the teacher's, we introduce a novel imitation distillation approach, which matches the policy's velocity to the teacher's along the policy's trajectory using a standard $\ell_2$ flow matching loss. By simply mimicking the teacher's behavior, $\pi$-Flow enables stable and scalable training and avoids the quality-diversity trade-off. On ImageNet 256$^2$, it attains a 1-NFE FID of 2.85, outperforming previous 1-NFE models of the same DiT architecture. On FLUX.1-12B and Qwen-Image-20B at 4 NFEs, $\pi$-Flow achieves substantially better diversity than state-of-the-art DMD models, while maintaining teacher-level quality.
Authors: Yongxian Wei, Yilin Zhao, Li Shen, Xinrui Chen, Runxi Cheng, Sinan Du, Hao Yu, Gang Liu, Jiahong Yan, Chun Yuan, Dian Li
Abstract: Data synthesis for training large reasoning models offers a scalable alternative to limited, human-curated datasets, enabling the creation of high-quality data. However, existing approaches face several challenges: (i) indiscriminate generation that ignores the solver's ability and yields low-value problems, or reliance on complex data pipelines to balance problem difficulty; and (ii) a lack of reasoning in problem generation, leading to shallow problem variants. In this paper, we develop a problem generator that reasons explicitly to plan problem directions before synthesis and adapts difficulty to the solver's ability. Specifically, we construct related problem pairs and augment them with intermediate problem-design CoT produced by a reasoning model. These data bootstrap problem-design strategies from the generator. Then, we treat the solver's feedback on synthetic problems as a reward signal, enabling the generator to calibrate difficulty and produce complementary problems near the edge of the solver's competence. Extensive experiments on 10 mathematical and general reasoning benchmarks show that our method achieves an average improvement of 2.5% and generalizes to both language and vision-language models. Moreover, a solver trained on the synthesized data provides improved rewards for continued generator training, enabling co-evolution and yielding a further 0.7% performance gain. Our code will be made publicly available here.
Authors: Mingyu Zhang, Lifeng Zhuo, Tianxi Tan, Guocan Xie, Xian Nie, Yan Li, Renjie Zhao, Zizhu He, Ziyu Wang, Jiting Cai, Yong-Lu Li
Abstract: Humans learn by observing, interacting with environments, and internalizing physics and causality. Here, we aim to ask whether an agent can similarly acquire human-like reasoning from interaction and keep improving with more experience. To study this, we introduce a Game-to-Unseen (G2U) benchmark of 1,000+ heterogeneous games that exhibit significant visual domain gaps. Existing approaches, including VLMs and world models, struggle to capture underlying physics and causality since they are not focused on core mechanisms and overfit to visual details. VLM/VLA agents reason but lack look-ahead in interactive settings, while world models imagine but imitate visual patterns rather than analyze physics and causality. We therefore propose IPR (Interactive Physical Reasoner), using world-model rollouts to score and reinforce a VLM's policy, and introduce PhysCode, a physics-centric action code aligning semantic intent with dynamics to provide a shared action space for prediction and reasoning. Pretrained on 1,000+ games, our IPR performs robustly on levels from primitive intuition to goal-driven reasoning, and even surpasses GPT-5 overall. We find that performance improves with more training games and interaction steps, and that the model also zero-shot transfers to unseen games. These results support physics-centric interaction as a path to steadily improving physical reasoning. Further demos and project details can be found at https://mybearyzhang.github.io/ipr-1.
Authors: Wei Zhen Teoh
Abstract: Jointly forecasting trajectories of multiple interacting agents is a core challenge in sports analytics and other domains involving complex group dynamics. Accurate prediction enables realistic simulation and strategic understanding of gameplay evolution. Most existing models are evaluated solely on per-agent accuracy metrics (minADE, minFDE), which assess each agent independently on its best-of-k prediction. However these metrics overlook whether the model learns which predicted trajectories can jointly form a plausible multi-agent future. Many state-of-the-art models are designed and optimized primarily based on these metrics. As a result, they may underperform on joint predictions and also fail to generate coherent, interpretable multi-agent scenarios in team sports. We propose CausalTraj, a temporally causal, likelihood-based model that is built to generate jointly probable multi-agent trajectory forecasts. To better assess collective modeling capability, we emphasize joint metrics (minJADE, minJFDE) that measure joint accuracy across agents within the best generated scenario sample. Evaluated on the NBA SportVU, Basketball-U, and Football-U datasets, CausalTraj achieves competitive per-agent accuracy and the best recorded results on joint metrics, while yielding qualitatively coherent and realistic gameplay evolutions.
Authors: Youssef Megahed, Saleh Abou-Alwan, Anthony Fuller, Dina El Demellawy, Steven Hawken, Adrian D. C. Chan
Abstract: Hirschsprung Disease is characterized by the absence of ganglion cells in the myenteric plexus. Therefore, the correct identification of ganglion cells is crucial for diagnosing Hirschsprung disease. We introduce a three-stage analysis framework that mimics the pathologist's diagnostic approach. The framework, based on a Vision Transformer model (ViT-B/16), sequentially segments the muscularis propria, segments the myenteric plexus, and detects ganglion cells within anatomically valid regions. 30 whole-slide images of colon tissue were used, each containing manual annotations of muscularis, plexus, and ganglion cells. A 5-fold cross-validation scheme was applied to each stage, along with resolution-specific tiling strategies and tailored postprocessing to ensure anatomical consistency. The proposed method achieved a Dice coefficient of 89.9% and a Plexus Inclusion Rate of 100% for muscularis segmentation. Plexus segmentation reached a recall of 94.8%, a precision of 84.2% and a Ganglia Inclusion Rate of 99.7%. For ganglion cells annotated with high certainty, the model achieved 62.1\% precision and 89.1% recall. When considering all annotated ganglion cells, regardless of certainty level, the overall precision was 67.0%. These results indicate that ViT-based models are effective at leveraging global tissue context and capturing cellular morphology at small scales, even within complex histological tissue structures. This multi-stage methodology has great potential to support digital pathology workflows by reducing inter-observer variability and assisting in the evaluation of Hirschsprung disease. The clinical impact will be evaluated in future work with larger multi-center datasets and additional expert annotations.
Authors: Jianlei Chang, Ruofeng Mei, Wei Ke, Xiangyu Xu
Abstract: Generative modeling has recently shown remarkable promise for visuomotor policy learning, enabling flexible and expressive control across diverse embodied AI tasks. However, existing generative policies often struggle with data inefficiency, requiring large-scale demonstrations, and sampling inefficiency, incurring slow action generation during inference. We introduce EfficientFlow, a unified framework for efficient embodied AI with flow-based policy learning. To enhance data efficiency, we bring equivariance into flow matching. We theoretically prove that when using an isotropic Gaussian prior and an equivariant velocity prediction network, the resulting action distribution remains equivariant, leading to improved generalization and substantially reduced data demands. To accelerate sampling, we propose a novel acceleration regularization strategy. As direct computation of acceleration is intractable for marginal flow trajectories, we derive a novel surrogate loss that enables stable and scalable training using only conditional trajectories. Across a wide range of robotic manipulation benchmarks, the proposed algorithm achieves competitive or superior performance under limited data while offering dramatically faster inference. These results highlight EfficientFlow as a powerful and efficient paradigm for high-performance embodied AI.
Authors: Gihyeon Lee, Jungwoo Lee, Juwon Kim, Young-Sik Shin, Younggun Cho
Abstract: Robots are often required to localize in environments with unknown object classes and semantic ambiguity. However, when performing global localization using semantic objects, high semantic ambiguity intensifies object misclassification and increases the likelihood of incorrect associations, which in turn can cause significant errors in the estimated pose. Thus, in this letter, we propose a multi-label likelihood-based semantic graph matching framework for object-level global localization. The key idea is to exploit multi-label graph representations, rather than single-label alternatives, to capture and leverage the inherent semantic context of object observations. Based on these representations, our approach enhances semantic correspondence across graphs by combining the likelihood of each node with the maximum likelihood of its neighbors via context-aware likelihood propagation. For rigorous validation, data association and pose estimation performance are evaluated under both closed-set and open-set detection configurations. In addition, we demonstrate the scalability of our approach to large-vocabulary object categories in both real-world indoor scenes and synthetic environments.
Authors: Haoran Jiang, Jin Chen, Qingwen Bu, Li Chen, Modi Shi, Yanjie Zhang, Delong Li, Chuanzhe Suo, Chuang Wang, Zhihui Peng, Hongyang Li
Abstract: Humanoid robots require precise locomotion and dexterous manipulation to perform challenging loco-manipulation tasks. Yet existing approaches, modular or end-to-end, are deficient in manipulation-aware locomotion. This confines the robot to a limited workspace, preventing it from performing large-space loco-manipulation. We attribute this to: (1) the challenge of acquiring loco-manipulation knowledge due to the scarcity of humanoid teleoperation data, and (2) the difficulty of faithfully and reliably executing locomotion commands, stemming from the limited precision and stability of existing RL controllers. To acquire richer loco-manipulation knowledge, we propose a unified latent learning framework that enables Vision-Language-Action (VLA) system to learn from low-cost action-free egocentric videos. Moreover, an efficient human data collection pipeline is devised to augment the dataset and scale the benefits. To execute the desired locomotion commands more precisely, we present a loco-manipulation-oriented (LMO) RL policy specifically tailored for accurate and stable core loco-manipulation movements, such as advancing, turning, and squatting. Building on these components, we introduce WholeBodyVLA, a unified framework for humanoid loco-manipulation. To the best of our knowledge, WholeBodyVLA is one of its kind enabling large-space humanoid loco-manipulation. It is verified via comprehensive experiments on the AgiBot X2 humanoid, outperforming prior baseline by 21.3%. It also demonstrates strong generalization and high extensibility across a broad range of tasks.