Authors: Yuncheng Lu, Yucen Shi, Aobo Li, Zehao Li, Junying Li, Bo Wang, Tony Tae-Hyoung Kim
Abstract: We present an energy-efficient anti-UAV system that integrates frame-based and event-driven object tracking to enable reliable detection of small and fast-moving drones. The system reconstructs binary event frames using run-length encoding, generates region proposals, and adaptively switches between frame mode and event mode based on object size and velocity. A Fast Object Tracking Unit improves robustness for high-speed targets through adaptive thresholding and trajectory-based classification. The neural processing unit supports both grayscale-patch and trajectory inference with a custom instruction set and a zero-skipping MAC architecture, reducing redundant neural computations by more than 97 percent. Implemented in 40 nm CMOS technology, the 2 mm^2 chip achieves 96 pJ per frame per pixel and 61 pJ per event at 0.8 V, and reaches 98.2 percent recognition accuracy on public UAV datasets across 50 to 400 m ranges and 5 to 80 pixels per second speeds. The results demonstrate state-of-the-art end-to-end energy efficiency for anti-UAV systems.
Authors: Karthik Prabhakar
Abstract: Nystagmus patients with photosensitivity face significant daily challenges due to involuntary eye movements exacerbated by environmental brightness conditions. Current assistive solutions are limited to symptomatic treatments without predictive personalization. This paper proposes NystagmusNet, an AI-driven system that predicts high-risk visual environments and recommends real-time visual adaptations. Using a dual-branch convolutional neural network trained on synthetic and augmented datasets, the system estimates a photosensitivity risk score based on environmental brightness and eye movement variance. The model achieves 75% validation accuracy on synthetic data. Explainability techniques including SHAP and GradCAM are integrated to highlight environmental risk zones, improving clinical trust and model interpretability. The system includes a rule-based recommendation engine for adaptive filter suggestions. Future directions include deployment via smart glasses and reinforcement learning for personalized recommendations.
Authors: Kaijie Chen, Zhiyang Xu, Ying Shen, Zihao Lin, Yuguang Yao, Lifu Huang
Abstract: Recent progress in flow-based generative models and reinforcement learning (RL) has improved text-image alignment and visual quality. However, current RL training for flow models still has two main problems: (i) GRPO-style fixed per-prompt group sizes ignore variation in sampling importance across prompts, which leads to inefficient sampling and slower training; and (ii) trajectory-level advantages are reused as per-step estimates, which biases credit assignment along the flow. We propose SuperFlow, an RL training framework for flow-based models that adjusts group sizes with variance-aware sampling and computes step-level advantages in a way that is consistent with continuous-time flow dynamics. Empirically, SuperFlow reaches promising performance while using only 5.4% to 56.3% of the original training steps and reduces training time by 5.2% to 16.7% without any architectural changes. On standard text-to-image (T2I) tasks, including text rendering, compositional image generation, and human preference alignment, SuperFlow improves over SD3.5-M by 4.6% to 47.2%, and over Flow-GRPO by 1.7% to 16.0%.
Authors: Ellie Zhou, Jihoon Chung, Olga Russakovsky
Abstract: Human action recognition models often rely on background cues rather than human movement and pose to make predictions, a behavior known as background bias. We present a systematic analysis of background bias across classification models, contrastive text-image pretrained models, and Video Large Language Models (VLLM) and find that all exhibit a strong tendency to default to background reasoning. Next, we propose mitigation strategies for classification models and show that incorporating segmented human input effectively decreases background bias by 3.78%. Finally, we explore manual and automated prompt tuning for VLLMs, demonstrating that prompt design can steer predictions towards human-focused reasoning by 9.85%.
Authors: Bin Wang, Fadi Dornaika
Abstract: Image classification is hindered by subtle inter-class differences and substantial intra-class variations, which limit the effectiveness of existing contrastive learning methods. Supervised contrastive approaches based on the InfoNCE loss suffer from negative-sample dilution and lack adaptive decision boundaries, thereby reducing discriminative power in fine-grained recognition tasks. To address these limitations, we propose Sigmoid-based Common and Style Supervised Contrastive Learning (SCS-SupCon). Our framework introduces a sigmoid-based pairwise contrastive loss with learnable temperature and bias parameters to enable adaptive decision boundaries. This formulation emphasizes hard negatives, mitigates negative-sample dilution, and more effectively exploits supervision. In addition, an explicit style-distance constraint further disentangles style and content representations, leading to more robust feature learning. Comprehensive experiments on six benchmark datasets, including CUB200-2011 and Stanford Dogs, demonstrate that SCS-SupCon achieves state-of-the-art performance across both CNN and Transformer backbones. On CIFAR-100 with ResNet-50, SCS-SupCon improves top-1 accuracy over SupCon by approximately 3.9 percentage points and over CS-SupCon by approximately 1.7 points under five-fold cross-validation. On fine-grained datasets, it outperforms CS-SupCon by 0.4--3.0 points. Extensive ablation studies and statistical analyses further confirm the robustness and generalization of the proposed framework, with Friedman tests and Nemenyi post-hoc evaluations validating the stability of the observed improvements.
Authors: Yuxiao Li
Abstract: We propose a modular framework for single-view indoor scene 3D reconstruction, where several core modules are powered by diffusion techniques. Traditional approaches for this task often struggle with the complex instance shapes and occlusions inherent in indoor environments. They frequently overshoot by attempting to predict 3D shapes directly from incomplete 2D images, which results in limited reconstruction quality. We aim to overcome this limitation by splitting the process into two steps: first, we employ diffusion-based techniques to predict the complete views of the room background and occluded indoor instances, then transform them into 3D. Our modular framework makes contributions to this field through the following components: an amodal completion module for restoring the full view of occluded instances, an inpainting model specifically trained to predict room layouts, a hybrid depth estimation technique that balances overall geometric accuracy with fine detail expressiveness, and a view-space alignment method that exploits both 2D and 3D cues to ensure precise placement of instances within the scene. This approach effectively reconstructs both foreground instances and the room background from a single image. Extensive experiments on the 3D-Front dataset demonstrate that our method outperforms current state-of-the-art (SOTA) approaches in terms of both visual quality and reconstruction accuracy. The framework holds promising potential for applications in interior design, real estate, and augmented reality.
Authors: Omar Faruq Shikdar, Fahad Ahammed, B. M. Shahria Alam, Golam Kibria, Tawhidur Rahman, Nishat Tasnim Niloy
Abstract: Tea is among the most widely consumed drinks globally. Tea production is a key industry for many countries. One of the main challenges in tea harvesting is tea leaf diseases. If the spread of tea leaf diseases is not stopped in time, it can lead to massive economic losses for farmers. Therefore, it is crucial to identify tea leaf diseases as soon as possible. Manually identifying tea leaf disease is an ineffective and time-consuming method, without any guarantee of success. Automating this process will improve both the efficiency and the success rate of identifying tea leaf diseases. The purpose of this study is to create an automated system that can classify different kinds of tea leaf diseases, allowing farmers to take action to minimize the damage. A novel dataset was developed specifically for this study. The dataset contains 5278 images across seven classes. The dataset was pre-processed prior to training the model. We deployed three pretrained models: DenseNet, Inception, and EfficientNet. EfficientNet was used only in the ensemble model. We utilized two different attention modules to improve model performance. The ensemble model achieved the highest accuracy of 85.68%. Explainable AI was introduced for better model interpretability.
Authors: Soumava Paul, Prakhar Kaushik, Ankit Vaidya, Anand Bhattad, Alan Yuille
Abstract: We address semantic 3D part segmentation: decomposing objects into parts with meaningful names. While datasets exist with part annotations, their definitions are inconsistent across datasets, limiting robust training. Previous methods produce unlabeled decompositions or retrieve single parts without complete shape annotations. We propose ALIGN-Parts, which formulates part naming as a direct set alignment task. Our method decomposes shapes into partlets - implicit 3D part representations - matched to part descriptions via bipartite assignment. We combine geometric cues from 3D part fields, appearance from multi-view vision features, and semantic knowledge from language-model-generated affordance descriptions. Text-alignment loss ensures partlets share embedding space with text, enabling a theoretically open-vocabulary matching setup, given sufficient data. Our efficient and novel, one-shot, 3D part segmentation and naming method finds applications in several downstream tasks, including serving as a scalable annotation engine. As our model supports zero-shot matching to arbitrary descriptions and confidence-calibrated predictions for known categories, with human verification, we create a unified ontology that aligns PartNet, 3DCoMPaT++, and Find3D, consisting of 1,794 unique 3D parts. We also show examples from our newly created Tex-Parts dataset. We also introduce 2 novel metrics appropriate for the named 3D part segmentation task.
Authors: Shubham Kumar Nigam, Parjanya Aditya Shukla, Noel Shallum, Arnab Bhattacharya
Abstract: Handwritten text recognition (HTR) and machine translation continue to pose significant challenges, particularly for low-resource languages like Marathi, which lack large digitized corpora and exhibit high variability in handwriting styles. The conventional approach to address this involves a two-stage pipeline: an OCR system extracts text from handwritten images, which is then translated into the target language using a machine translation model. In this work, we explore and compare the performance of traditional OCR-MT pipelines with Vision Large Language Models that aim to unify these stages and directly translate handwritten text images in a single, end-to-end step. Our motivation is grounded in the urgent need for scalable, accurate translation systems to digitize legal records such as FIRs, charge sheets, and witness statements in India's district and high courts. We evaluate both approaches on a curated dataset of handwritten Marathi legal documents, with the goal of enabling efficient legal document processing, even in low-resource environments. Our findings offer actionable insights toward building robust, edge-deployable solutions that enhance access to legal information for non-native speakers and legal professionals alike.
Authors: Fakrul Islam Tushar, Ehsan Samei, Cynthia Rudin, Joseph Y. Lo
Abstract: Objective: Although medical imaging datasets are increasingly available, abnormal and annotation-intensive findings critical to lung cancer screening, particularly small pulmonary nodules, remain underrepresented and inconsistently curated. Methods: We introduce NodMAISI, an anatomically constrained, nodule-oriented CT synthesis and augmentation framework trained on a unified multi-source cohort (7,042 patients, 8,841 CTs, 14,444 nodules). The framework integrates: (i) a standardized curation and annotation pipeline linking each CT with organ masks and nodule-level annotations, (ii) a ControlNet-conditioned rectified-flow generator built on MAISI-v2's foundational blocks to enforce anatomy- and lesion-consistent synthesis, and (iii) lesion-aware augmentation that perturbs nodule masks (controlled shrinkage) while preserving surrounding anatomy to generate paired CT variants. Results: Across six public test datasets, NodMAISI improved distributional fidelity relative to MAISI-v2 (real-to-synthetic FID range 1.18 to 2.99 vs 1.69 to 5.21). In lesion detectability analysis using a MONAI nodule detector, NodMAISI substantially increased average sensitivity and more closely matched clinical scans (IMD-CT: 0.69 vs 0.39; DLCS24: 0.63 vs 0.20), with the largest gains for sub-centimeter nodules where MAISI-v2 frequently failed to reproduce the conditioned lesion. In downstream nodule-level malignancy classification trained on LUNA25 and externally evaluated on LUNA16, LNDbv4, and DLCS24, NodMAISI augmentation improved AUC by 0.07 to 0.11 at <=20% clinical data and by 0.12 to 0.21 at 10%, consistently narrowing the performance gap under data scarcity.
Authors: Ami Pandat, Punna Rajasekhar, Gopika Vinod, Rohit Shukla
Abstract: Unmanned Aerial Vehicles, commonly known as, drones pose increasing risks in civilian and defense settings, demanding accurate and real-time drone detection systems. However, detecting drones is challenging because of their small size, rapid movement, and low visual contrast. A modified architecture of YolovN called the YolovN-CBi is proposed that incorporates the Convolutional Block Attention Module (CBAM) and the Bidirectional Feature Pyramid Network (BiFPN) to improve sensitivity to small object detections. A curated training dataset consisting of 28K images is created with various flying objects and a local test dataset is collected with 2500 images consisting of very small drone objects. The proposed architecture is evaluated on four benchmark datasets, along with the local test dataset. The baseline Yolov5 and the proposed Yolov5-CBi architecture outperform newer Yolo versions, including Yolov8 and Yolov12, in the speed-accuracy trade-off for small object detection. Four other variants of the proposed CBi architecture are also proposed and evaluated, which vary in the placement and usage of CBAM and BiFPN. These variants are further distilled using knowledge distillation techniques for edge deployment, using a Yolov5m-CBi teacher and a Yolov5n-CBi student. The distilled model achieved a mA@P0.5:0.9 of 0.6573, representing a 6.51% improvement over the teacher's score of 0.6171, highlighting the effectiveness of the distillation process. The distilled model is 82.9% faster than the baseline model, making it more suitable for real-time drone detection. These findings highlight the effectiveness of the proposed CBi architecture, together with the distilled lightweight models in advancing efficient and accurate real-time detection of small UAVs.
Authors: Sabri Mustafa Kahya, Muhammet Sami Yavuz, Boran Hamdi Sivrikaya, Eckehard Steinbach
Abstract: Out-of-distribution (OOD) detection is essential for the safe deployment of neural networks, as it enables the identification of samples outside the training domain. We present FOODER, a real-time, privacy-preserving radar-based framework that integrates OOD-based facial authentication with facial expression recognition. FOODER operates using low-cost frequency-modulated continuous-wave (FMCW) radar and exploits both range-Doppler and micro range-Doppler representations. The authentication module employs a multi-encoder multi-decoder architecture with Body Part (BP) and Intermediate Linear Encoder-Decoder (ILED) components to classify a single enrolled individual as in-distribution while detecting all other faces as OOD. Upon successful authentication, an expression recognition module is activated. Concatenated radar representations are processed by a ResNet block to distinguish between dynamic and static facial expressions. Based on this categorization, two specialized MobileViT networks are used to classify dynamic expressions (smile, shock) and static expressions (neutral, anger). This hierarchical design enables robust facial authentication and fine-grained expression recognition while preserving user privacy by relying exclusively on radar data. Experiments conducted on a dataset collected with a 60 GHz short-range FMCW radar demonstrate that FOODER achieves an AUROC of 94.13% and an FPR95 of 18.12% for authentication, along with an average expression recognition accuracy of 94.70%. FOODER outperforms state-of-the-art OOD detection methods and several transformer-based architectures while operating efficiently in real time.
Authors: Ekta Balkrishna Gavas, Sudipta Banerjee, Chinmay Hegde, Nasir Memon
Abstract: Multimodal LLMs (MLLMs) have gained significant traction in complex data analysis, visual question answering, generation, and reasoning. Recently, they have been used for analyzing the biometric utility of iris and face images. However, their capabilities in fingerprint understanding are yet unexplored. In this work, we design a comprehensive benchmark, \textsc{FPBench} that evaluates the performance of 20 MLLMs (open-source and proprietary) across 7 real and synthetic datasets on 8 biometric and forensic tasks using zero-shot and chain-of-thought prompting strategies. We discuss our findings in terms of performance, explainability and share our insights into the challenges and limitations. We establish \textsc{FPBench} as the first comprehensive benchmark for fingerprint domain understanding with MLLMs paving the path for foundation models for fingerprints.
Authors: Shreshth Rajan, Raymond Liu
Abstract: Semantic segmentation of outdoor street scenes plays a key role in applications such as autonomous driving, mobile robotics, and assistive technology for visually-impaired pedestrians. For these applications, accurately distinguishing between key surfaces and objects such as roads, sidewalks, vehicles, and pedestrians is essential for maintaining safety and minimizing risks. Semantic segmentation must be robust to different environments, lighting and weather conditions, and sensor noise, while being performed in real-time. We propose a region-level, uncertainty-gated retrieval mechanism that improves segmentation accuracy and calibration under domain shift. Our best method achieves an 11.3% increase in mean intersection-over-union while reducing retrieval cost by 87.5%, retrieving for only 12.5% of regions compared to 100% for always-on baseline.
Authors: Thomas Boudras, Martin Schwartz, Rasmus Fensholt, Martin Brandt, Ibrahim Fayad, Jean-Pierre Wigneron, Gabriel Belouze, Fajwel Fogel, Philippe Ciais
Abstract: High-resolution mapping of canopy height is essential for forest management and biodiversity monitoring. Although recent studies have led to the advent of deep learning methods using satellite imagery to predict height maps, these approaches often face a trade-off between data accessibility and spatial resolution. To overcome these limitations, we present SERA-H, an end-to-end model combining a super-resolution module (EDSR) and temporal attention encoding (UTAE). Trained under the supervision of high-density LiDAR data (ALS), our model generates 2.5 m resolution height maps from freely available Sentinel-1 and Sentinel-2 (10 m) time series data. Evaluated on an open-source benchmark dataset in France, SERA-H, with a MAE of 2.6 m and a coefficient of determination of 0.82, not only outperforms standard Sentinel-1/2 baselines but also achieves performance comparable to or better than methods relying on commercial very high-resolution imagery (SPOT-6/7, PlanetScope, Maxar). These results demonstrate that combining high-resolution supervision with the spatiotemporal information embedded in time series enables the reconstruction of details beyond the input sensors' native resolution. SERA-H opens the possibility of freely mapping forests with high revisit frequency, achieving accuracy comparable to that of costly commercial imagery. The source code is available at https://github.com/ThomasBoudras/SERA-H#
Authors: Hao Li, Daiwei Lu, Jiacheng Wang, Robert J. Webster III, Ipek Oguz
Abstract: This work presents EndoStreamDepth, a monocular depth estimation framework for endoscopic video streams. It provides accurate depth maps with sharp anatomical boundaries for each frame, temporally consistent predictions across frames, and real-time throughput. Unlike prior work that uses batched inputs, EndoStreamDepth processes individual frames with a temporal module to propagate inter-frame information. The framework contains three main components: (1) a single-frame depth network with endoscopy-specific transformation to produce accurate depth maps, (2) multi-level Mamba temporal modules that leverage inter-frame information to improve accuracy and stabilize predictions, and (3) a hierarchical design with comprehensive multi-scale supervision, where complementary loss terms jointly improve local boundary sharpness and global geometric consistency. We conduct comprehensive evaluations on two publicly available colonoscopy depth estimation datasets. Compared to state-of-the-art monocular depth estimation methods, EndoStreamDepth substantially improves performance, and it produces depth maps with sharp, anatomically aligned boundaries, which are essential to support downstream tasks such as automation for robotic surgery. The code is publicly available at https://github.com/MedICL-VU/EndoStreamDepth
Authors: Taewon Yang, Jason Hu, Jeffrey A. Fessler, Liyue Shen
Abstract: Diffusion models learn strong image priors that can be leveraged to solve inverse problems like medical image reconstruction. However, for real-world applications such as 3D Computed Tomography (CT) imaging, directly training diffusion models on 3D data presents significant challenges due to the high computational demands of extensive GPU resources and large-scale datasets. Existing works mostly reuse 2D diffusion priors to address 3D inverse problems, but fail to fully realize and leverage the generative capacity of diffusion models for high-dimensional data. In this study, we propose a novel 3D patch-based diffusion model that can learn a fully 3D diffusion prior from limited data, enabling scalable generation of high-resolution 3D images. Our core idea is to learn the prior of 3D patches to achieve scalable efficiency, while coupling local and global information to guarantee high-quality 3D image generation, by modeling the joint distribution of position-aware 3D local patches and downsampled 3D volume as global context. Our approach not only enables high-quality 3D generation, but also offers an unprecedentedly efficient and accurate solution to high-resolution 3D inverse problems. Experiments on 3D CT reconstruction across multiple datasets show that our method outperforms state-of-the-art methods in both performance and efficiency, notably achieving high-resolution 3D reconstruction of $512 \times 512 \times 256$ ($\sim$20 mins).
Authors: Ziyu Zhang, Yi Yu, Simeng Zhu, Ahmed Aly, Yunhe Gao, Ning Gu, Yuan Xue
Abstract: Accurate medical image segmentation is essential for clinical diagnosis and treatment planning. While recent interactive foundation models (e.g., nnInteractive) enhance generalization through large-scale multimodal pretraining, they still depend on precise prompts and often perform below expectations in contexts that are underrepresented in their training data. We present AtlasSegFM, an atlas-guided framework that customizes available foundation models to clinical contexts with a single annotated example. The core innovations are: 1) a pipeline that provides context-aware prompts for foundation models via registration between a context atlas and query images, and 2) a test-time adapter to fuse predictions from both atlas registration and the foundation model. Extensive experiments across public and in-house datasets spanning multiple modalities and organs demonstrate that AtlasSegFM consistently improves segmentation, particularly for small, delicate structures. AtlasSegFM provides a lightweight, deployable solution one-shot customization of foundation models in real-world clinical workflows. The code will be made publicly available.
Authors: Kaixing Yang, Jiashu Zhu, Xulong Tang, Ziqiao Peng, Xiangyue Zhang, Puwei Wang, Jiahong Wu, Xiangxiang Chu, Hongyan Liu, Jun He
Abstract: With the rise of online dance-video platforms and rapid advances in AI-generated content (AIGC), music-driven dance generation has emerged as a compelling research direction. Despite substantial progress in related domains such as music-driven 3D dance generation, pose-driven image animation, and audio-driven talking-head synthesis, existing methods cannot be directly adapted to this task. Moreover, the limited studies in this area still struggle to jointly achieve high-quality visual appearance and realistic human motion. Accordingly, we present MACE-Dance, a music-driven dance video generation framework with cascaded Mixture-of-Experts (MoE). The Motion Expert performs music-to-3D motion generation while enforcing kinematic plausibility and artistic expressiveness, whereas the Appearance Expert carries out motion- and reference-conditioned video synthesis, preserving visual identity with spatiotemporal coherence. Specifically, the Motion Expert adopts a diffusion model with a BiMamba-Transformer hybrid architecture and a Guidance-Free Training (GFT) strategy, achieving state-of-the-art (SOTA) performance in 3D dance generation. The Appearance Expert employs a decoupled kinematic-aesthetic fine-tuning strategy, achieving state-of-the-art (SOTA) performance in pose-driven image animation. To better benchmark this task, we curate a large-scale and diverse dataset and design a motion-appearance evaluation protocol. Based on this protocol, MACE-Dance also achieves state-of-the-art performance. Project page: https://macedance.github.io/
Authors: Junho Lee, Kwanseok Kim, Joonseok Lee
Abstract: Flow matching has emerged as a powerful generative modeling approach with flexible choices of source distribution. While Gaussian distributions are commonly used, the potential for better alternatives in high-dimensional data generation remains largely unexplored. In this paper, we propose a novel 2D simulation that captures high-dimensional geometric properties in an interpretable 2D setting, enabling us to analyze the learning dynamics of flow matching during training. Based on this analysis, we derive several key insights about flow matching behavior: (1) density approximation can paradoxically degrade performance due to mode discrepancy, (2) directional alignment suffers from path entanglement when overly concentrated, (3) Gaussian's omnidirectional coverage ensures robust learning, and (4) norm misalignment incurs substantial learning costs. Building on these insights, we propose a practical framework that combines norm-aligned training with directionally-pruned sampling. This approach maintains the robust omnidirectional supervision essential for stable flow learning, while eliminating initializations in data-sparse regions during inference. Importantly, our pruning strategy can be applied to any flow matching model trained with a Gaussian source, providing immediate performance gains without the need for retraining. Empirical evaluations demonstrate consistent improvements in both generation quality and sampling efficiency. Our findings provide practical insights and guidelines for source distribution design and introduce a readily applicable technique for improving existing flow matching models. Our code is available at https://github.com/kwanseokk/SourceFM.
Authors: Janghyun Baek, Mincheol Chang, Seokha Moon, Seung Joon Lee, Jinkyu Kim
Abstract: Recent query-based 3D object detection methods using camera and LiDAR inputs have shown strong performance, but existing query initialization strategies,such as random sampling or BEV heatmap-based sampling, often result in inefficient query usage and reduced accuracy, particularly for occluded or crowded objects. To address this limitation, we propose ALIGN (Advanced query initialization with LiDAR and Image GuidaNce), a novel approach for occlusion-robust, object-aware query initialization. Our model consists of three key components: (i) Occlusion-aware Center Estimation (OCE), which integrates LiDAR geometry and image semantics to estimate object centers accurately (ii) Adaptive Neighbor Sampling (ANS), which generates object candidates from LiDAR clustering and supplements each object by sampling spatially and semantically aligned points around it and (iii) Dynamic Query Balancing (DQB), which adaptively balances queries between foreground and background regions. Our extensive experiments on the nuScenes benchmark demonstrate that ALIGN consistently improves performance across multiple state-of-the-art detectors, achieving gains of up to +0.9 mAP and +1.2 NDS, particularly in challenging scenes with occlusions or dense crowds. Our code will be publicly available upon publication.
Authors: Alex Foo, Wynne Hsu, Mong Li Lee
Abstract: Discovering object-centric representations from images can significantly enhance the robustness, sample efficiency and generalizability of vision models. Works on images with multi-part objects typically follow an implicit object representation approach, which fail to recognize these learned objects in occluded or out-of-distribution contexts. This is due to the assumption that object part-whole relations are implicitly encoded into the representations through indirect training objectives. We address this limitation by proposing a novel method that leverages on explicit graph representations for parts and present a co-part object discovery algorithm. We then introduce three benchmarks to evaluate the robustness of object-centric methods in recognizing multi-part objects within occluded and out-of-distribution settings. Experimental results on simulated, realistic, and real-world images show marked improvements in the quality of discovered objects compared to state-of-the-art methods, as well as the accurate recognition of multi-part objects in occluded and out-of-distribution contexts. We also show that the discovered object-centric representations can more accurately predict key object properties in a downstream task, highlighting the potential of our method to advance the field of object-centric representations.
Authors: Mohammad Zolfaghari, Hedieh Sajedi
Abstract: Anomaly detection or outlier is one of the challenging subjects in unsupervised learning . This paper is introduced a student-teacher framework for anomaly detection that its teacher network is enhanced for achieving high-performance metrics . For this purpose , we first pre-train the ResNet-18 network on the ImageNet and then fine-tune it on the MVTech-AD dataset . Experiment results on the image-level and pixel-level demonstrate that this idea has achieved better metrics than the previous methods . Our model , Enhanced Teacher for Student-Teacher Feature Pyramid (ET-STPM), achieved 0.971 mean accuracy on the image-level and 0.977 mean accuracy on the pixel-level for anomaly detection.
Authors: Takuya OKi, Yuan Liu
Abstract: Understanding spatial openness is vital for improving residential quality and design; however, studies often treat its influencing factors separately. This study developed a quantitative framework to evaluate the spatial openness in housing from two- (2D) and three- (3D) dimensional perspectives. Using data from 4,004 rental units in Tokyo's 23 wards, we examined the temporal and spatial variations in openness and its relationship with rent and housing attributes. 2D openness was computed via planar visibility using visibility graph analysis (VGA) from floor plans, whereas 3D openness was derived from interior images analysed using Mask2Former, a semantic segmentation model that identifies walls, ceilings, floors, and windows. The results showed an increase in living room visibility and a 1990s peak in overall openness. Spatial analyses revealed partial correlations among openness, rent, and building characteristics, reflecting urban redevelopment trends. Although the 2D and 3D openness indicators were not directly correlated, higher openness tended to correspond to higher rent. The impression scores predicted by the existing models were only weakly related to openness, suggesting that the interior design and furniture more strongly shape perceived space. This study offers a new multidimensional data-driven framework for quantifying residential spatial openness and linking it with urban and market dynamics.
Authors: Aryan Chaudhary, Sanchit Goyal, Pratik Narang, Dhruv Kumar
Abstract: Vision-Language Models have demonstrated remarkable capabilities in understanding visual content, yet systematic biases in their spatial processing remain largely unexplored. This work identifies and characterizes a systematic spatial attention bias where VLMs consistently prioritize describing left-positioned content before right-positioned content in horizontally concatenated images. Through controlled experiments on image pairs using both open-source and closed-source models, we demonstrate that this bias persists across different architectures, with models describing left-positioned content first in approximately 97% of cases under neutral prompting conditions. Testing on an Arabic-finetuned model reveals that the bias persists despite right-to-left language training, ruling out language reading direction as the primary cause. Investigation of training dataset annotation guidelines from PixMo and Visual Genome reveals no explicit left-first ordering instructions, suggesting the bias is consistent with architectural factors rather than explicit training data instructions. These findings reveal fundamental limitations in how current VLMs process spatial information.
Authors: Shahram Najam Syed, Yitian Hu, Yuchao Yao
Abstract: Photorealistic 3-D reconstruction from monocular video collapses in large-scale scenes when depth, pose, and radiance are solved in isolation: scale-ambiguous depth yields ghost geometry, long-horizon pose drift corrupts alignment, and a single global NeRF cannot model hundreds of metres of content. We introduce a joint learning framework that couples all three factors and demonstrably overcomes each failure case. Our system begins with a Vision-Transformer (ViT) depth network trained with metric-scale supervision, giving globally consistent depths despite wide field-of-view variations. A multi-scale feature bundle-adjustment (BA) layer refines camera poses directly in feature space--leveraging learned pyramidal descriptors instead of brittle keypoints--to suppress drift on unconstrained trajectories. For scene representation, we deploy an incremental local-radiance-field hierarchy: new hash-grid NeRFs are allocated and frozen on-the-fly when view overlap falls below a threshold, enabling city-block-scale coverage on a single GPU. Evaluated on the Tanks and Temples benchmark, our method reduces Absolute Trajectory Error to 0.001-0.021 m across eight indoor-outdoor sequences--up to 18x lower than BARF and 2x lower than NoPe-NeRF--while maintaining sub-pixel Relative Pose Error. These results demonstrate that metric-scale, drift-free 3-D reconstruction and high-fidelity novel-view synthesis are achievable from a single uncalibrated RGB camera.
Authors: Pan Ben Wong, Chengli Wu, Hanyue Lu
Abstract: Real-time Video Frame Interpolation (VFI) has long been dominated by flow-based methods like RIFE, which offer high throughput but often fail in complicated scenarios involving large motion and occlusion. Conversely, recent diffusion-based approaches (e.g., Consec. BB) achieve state-of-the-art perceptual quality but suffer from prohibitive latency, rendering them impractical for real-time applications. To bridge this gap, we propose Semantic-Guided RIFE (SG-RIFE). Instead of training from scratch, we introduce a parameter-efficient fine-tuning strategy that augments a pre-trained RIFE backbone with semantic priors from a frozen DINOv3 Vision Transformer. We propose a Split-Fidelity Aware Projection Module (Split-FAPM) to compress and refine high-dimensional features, and a Deformable Semantic Fusion (DSF) module to align these semantic priors with pixel-level motion fields. Experiments on SNU-FILM demonstrate that semantic injection provides a decisive boost in perceptual fidelity. SG-RIFE outperforms diffusion-based LDMVFI in FID/LPIPS and achieves quality comparable to Consec. BB on complex benchmarks while running significantly faster, proving that semantic consistency enables flow-based methods to achieve diffusion-competitive perceptual quality in near real-time.
Authors: Xiao He, Chang Tang, Xinwang Liu, Wei Zhang, Zhimin Gao, Chuankun Li, Shaohua Qiu, Jiangfeng Xu
Abstract: Hyperspectral images with high spectral resolution provide new insights into recognizing subtle differences in similar substances. However, object detection in hyperspectral images faces significant challenges in intra- and inter-class similarity due to the spatial differences in hyperspectral inter-bands and unavoidable interferences, e.g., sensor noises and illumination. To alleviate the hyperspectral inter-bands inconsistencies and redundancy, we propose a novel network termed \textbf{S}pectral \textbf{D}iscrepancy and \textbf{C}ross-\textbf{M}odal semantic consistency learning (SDCM), which facilitates the extraction of consistent information across a wide range of hyperspectral bands while utilizing the spectral dimension to pinpoint regions of interest. Specifically, we leverage a semantic consistency learning (SCL) module that utilizes inter-band contextual cues to diminish the heterogeneity of information among bands, yielding highly coherent spectral dimension representations. On the other hand, we incorporate a spectral gated generator (SGG) into the framework that filters out the redundant data inherent in hyperspectral information based on the importance of the bands. Then, we design the spectral discrepancy aware (SDA) module to enrich the semantic representation of high-level information by extracting pixel-level spectral features. Extensive experiments on two hyperspectral datasets demonstrate that our proposed method achieves state-of-the-art performance when compared with other ones.
Authors: Rui Xing, Runmin Cong, Yingying Wu, Can Wang, Zhongming Tang, Fen Wang, Hao Wu, Sam Kwong
Abstract: Understanding the dietary preferences of ancient societies and their evolution across periods and regions is crucial for revealing human-environment interactions. Seeds, as important archaeological artifacts, represent a fundamental subject of archaeobotanical research. However, traditional studies rely heavily on expert knowledge, which is often time-consuming and inefficient. Intelligent analysis methods have made progress in various fields of archaeology, but there remains a research gap in data and methods in archaeobotany, especially in the classification task of ancient plant seeds. To address this, we construct the first Ancient Plant Seed Image Classification (APS) dataset. It contains 8,340 images from 17 genus- or species-level seed categories excavated from 18 archaeological sites across China. In addition, we design a framework specifically for the ancient plant seed classification task (APSNet), which introduces the scale feature (size) of seeds based on learning fine-grained information to guide the network in discovering key "evidence" for sufficient classification. Specifically, we design a Size Perception and Embedding (SPE) module in the encoder part to explicitly extract size information for the purpose of complementing fine-grained information. We propose an Asynchronous Decoupled Decoding (ADD) architecture based on traditional progressive learning to decode features from both channel and spatial perspectives, enabling efficient learning of discriminative features. In both quantitative and qualitative analyses, our approach surpasses existing state-of-the-art image classification methods, achieving an accuracy of 90.5%. This demonstrates that our work provides an effective tool for large-scale, systematic archaeological research.
Authors: Mingcheng Ye, Jiaming Liu, Yiren Song
Abstract: Interleaved text-image generation aims to jointly produce coherent visual frames and aligned textual descriptions within a single sequence, enabling tasks such as style transfer, compositional synthesis, and procedural tutorials. We present Loom, a unified diffusion-transformer framework for interleaved text-image generation. Loom extends the Bagel unified model via full-parameter fine-tuning and an interleaved architecture that alternates textual and visual embeddings for multi-condition reasoning and sequential planning. A language planning strategy first decomposes a user instruction into stepwise prompts and frame embeddings, which guide temporally consistent synthesis. For each frame, Loom conditions on a small set of sampled prior frames together with the global textual context, rather than concatenating all history, yielding controllable and efficient long-horizon generation. Across style transfer, compositional generation, and tutorial-like procedures, Loom delivers superior compositionality, temporal coherence, and text-image alignment. Experiments demonstrate that Loom substantially outperforms the open-source baseline Anole, achieving an average gain of 2.6 points (on a 5-point scale) across temporal and semantic metrics in text-to-interleaved tasks. We also curate a 50K interleaved tutorial dataset and demonstrate strong improvements over unified and diffusion editing baselines.
Authors: Yucheng Fan, Jiawei Chen, Yu Tian, Zhaoxia Yin
Abstract: As vision-language models (VLMs) become widely adopted, VLM-based attribute inference attacks have emerged as a serious privacy concern, enabling adversaries to infer private attributes from images shared on social media. This escalating threat calls for dedicated protection methods to safeguard user privacy. However, existing methods often degrade the visual quality of images or interfere with vision-based functions on social media, thereby failing to achieve a desirable balance between privacy protection and user experience. To address this challenge, we propose a novel protection method that jointly optimizes privacy suppression and utility preservation under a visual consistency constraint. While our method is conceptually effective, fair comparisons between methods remain challenging due to the lack of publicly available evaluation datasets. To fill this gap, we introduce VPI-COCO, a publicly available benchmark comprising 522 images with hierarchically structured privacy questions and corresponding non-private counterparts, enabling fine-grained and joint evaluation of protection methods in terms of privacy preservation and user experience. Building upon this benchmark, experiments on multiple VLMs demonstrate that our method effectively reduces PAR below 25%, keeps NPAR above 88%, maintains high visual consistency, and generalizes well to unseen and paraphrased privacy questions, demonstrating its strong practical applicability for real-world VLM deployments.
Authors: Se-Young Jang, Su-Yeon Yoon, Jae-Woong Jung, Dong-Hun Lee, Seong-Hun Choi, Soo-Kyung Jun, Yu-Bin Kim, Young-Seon Ju, Kyounggon Kim
Abstract: With the accelerating pace of digital transformation and the widespread adoption of online platforms, both social and technical concerns regarding dark patterns-user interface designs that undermine users' ability to make informed and rational choices-have become increasingly prominent. As corporate online platforms grow more sophisticated in their design strategies, there is a pressing need for proactive and real-time detection technologies that go beyond the predominantly reactive approaches employed by regulatory authorities. In this paper, we propose a visual dark pattern detection framework that improves both detection accuracy and real-time performance. To this end, we constructed a proprietary visual object detection dataset by manually collecting 4,066 UI/UX screenshots containing dark patterns from 194 websites across six major industrial sectors in South Korea and abroad. The collected images were annotated with five representative UI components commonly associated with dark patterns: Button, Checkbox, Input Field, Pop-up, and QR Code. This dataset has been publicly released to support further research and development in the field. To enable real-time detection, this study adopted the YOLOv12x object detection model and applied transfer learning to optimize its performance for visual dark pattern recognition. Experimental results demonstrate that the proposed approach achieves a high detection accuracy of 92.8% in terms of mAP@50, while maintaining a real-time inference speed of 40.5 frames per second (FPS), confirming its effectiveness for practical deployment in online environments. Furthermore, to facilitate future research and contribute to technological advancements, the dataset constructed in this study has been made publicly available at https://github.com/B4E2/B4E2-DarkPattern-YOLO-DataSet.
URLs: https://github.com/B4E2/B4E2-DarkPattern-YOLO-DataSet.
Authors: Zhangshuo Qi, Jingyi Xu, Luqi Cheng, Shichen Wen, Yiming Ma, Guangming Xiong
Abstract: Place recognition is a critical component of autonomous vehicles and robotics, enabling global localization in GPS-denied environments. Recent advances have spurred significant interest in multimodal place recognition (MPR), which leverages complementary strengths of multiple modalities. Despite its potential, most existing MPR methods still face three key challenges: (1) dynamically adapting to arbitrary modality inputs within a unified framework, (2) maintaining robustness with missing or degraded modalities, and (3) generalizing across diverse sensor configurations and setups. In this paper, we propose UniMPR, a unified framework for multimodal place recognition. Using only one trained model, it can seamlessly adapt to any combination of common perceptual modalities (e.g., camera, LiDAR, radar). To tackle the data heterogeneity, we unify all inputs within a polar BEV feature space. Subsequently, the polar BEVs are fed into a multi-branch network to exploit discriminative intra-model and inter-modal features from any modality combinations. To fully exploit the network's generalization capability and robustness, we construct a large-scale training set from multiple datasets and introduce an adaptive label assignment strategy for extensive pre-training. Experiments on seven datasets demonstrate that UniMPR achieves state-of-the-art performance under varying sensor configurations, modality combinations, and environmental conditions. Our code will be released at https://github.com/QiZS-BIT/UniMPR.
Authors: Zidong Gu, Shoufu Tian
Abstract: Object detection in aerial imagery is a critical task in applications such as UAV reconnaissance. Although existing methods have extensively explored feature interaction between different modalities, they commonly rely on simple fusion strategies for feature aggregation. This introduces two critical flaws: it is prone to cross-modal noise and disrupts the hierarchical structure of the feature pyramid, thereby impairing the fine-grained detection of small objects. To address this challenge, we propose the Pyramidal Adaptive Cross-Gating Network (PACGNet), an architecture designed to perform deep fusion within the backbone. To this end, we design two core components: the Symmetrical Cross-Gating (SCG) module and the Pyramidal Feature-aware Multimodal Gating (PFMG) module. The SCG module employs a bidirectional, symmetrical "horizontal" gating mechanism to selectively absorb complementary information, suppress noise, and preserve the semantic integrity of each modality. The PFMG module reconstructs the feature hierarchy via a progressive hierarchical gating mechanism. This leverages the detailed features from a preceding, higher-resolution level to guide the fusion at the current, lower-resolution level, effectively preserving fine-grained details as features propagate. Through evaluations conducted on the DroneVehicle and VEDAI datasets, our PACGNet sets a new state-of-the-art benchmark, with mAP50 scores reaching 81.7% and 82.1% respectively.
Authors: Zeyu Zhang, Wei Zhai, Jian Yang, Yang Cao
Abstract: The creation of high-fidelity, physically-based rendering (PBR) materials remains a bottleneck in many graphics pipelines, typically requiring specialized equipment and expert-driven post-processing. To democratize this process, we present MatE, a novel method for generating tileable PBR materials from a single image taken under unconstrained, real-world conditions. Given an image and a user-provided mask, MatE first performs coarse rectification using an estimated depth map as a geometric prior, and then employs a dual-branch diffusion model. Leveraging a learned consistency from rotation-aligned and scale-aligned training data, this model further rectify residual distortions from the coarse result and translate it into a complete set of material maps, including albedo, normal, roughness and height. Our framework achieves invariance to the unknown illumination and perspective of the input image, allowing for the recovery of intrinsic material properties from casual captures. Through comprehensive experiments on both synthetic and real-world data, we demonstrate the efficacy and robustness of our approach, enabling users to create realistic materials from real-world image.
Authors: Philipp Langsteiner, Jan-Niklas Dihlmann, Hendrik P. A. Lensch
Abstract: Manual modeling of material parameters and 3D geometry is a time consuming yet essential task in the gaming and film industries. While recent advances in 3D reconstruction have enabled accurate approximations of scene geometry and appearance, these methods often fall short in relighting scenarios due to the lack of precise, spatially varying material parameters. At the same time, diffusion models operating on 2D images have shown strong performance in predicting physically based rendering (PBR) properties such as albedo, roughness, and metallicity. However, transferring these 2D material maps onto reconstructed 3D geometry remains a significant challenge. We propose a framework for fusing 2D material data into 3D geometry using a combination of novel learning-based and projection-based approaches. We begin by reconstructing scene geometry via Gaussian Splatting. From the input images, a diffusion model generates 2D maps for albedo, roughness, and metallic parameters. Any existing diffusion model that can convert images or videos to PBR materials can be applied. The predictions are further integrated into the 3D representation either by optimizing an image-based loss or by directly projecting the material parameters onto the Gaussians using Gaussian ray tracing. To enhance fine-scale accuracy and multi-view consistency, we further introduce a light-weight neural refinement step (Neural Merger), which takes ray-traced material features as input and produces detailed adjustments. Our results demonstrate that the proposed methods outperform existing techniques in both quantitative metrics and perceived visual realism. This enables more accurate, relightable, and photorealistic renderings from reconstructed scenes, significantly improving the realism and efficiency of asset creation workflows in content production pipelines.
Authors: Qiong Lou, Han Yang, Fang Lu
Abstract: Bone Age Assessment (BAA) is a widely used clinical technique that can accurately reflect an individual's growth and development level, as well as maturity. In recent years, although deep learning has advanced the field of bone age assessment, existing methods face challenges in efficiently balancing global features and local skeletal details. This study aims to develop an automated bone age assessment system based on a two-stream deep learning architecture to achieve higher accuracy in bone age assessment. We propose the BoNet+ model incorporating global and local feature extraction channels. A Transformer module is introduced into the global feature extraction channel to enhance the ability in extracting global features through multi-head self-attention mechanism. A RFAConv module is incorporated into the local feature extraction channel to generate adaptive attention maps within multiscale receptive fields, enhancing local feature extraction capabilities. Global and local features are concatenated along the channel dimension and optimized by an Inception-V3 network. The proposed method has been validated on the Radiological Society of North America (RSNA) and Radiological Hand Pose Estimation (RHPE) test datasets, achieving mean absolute errors (MAEs) of 3.81 and 5.65 months, respectively. These results are comparable to the state-of-the-art. The BoNet+ model reduces the clinical workload and achieves automatic, high-precision, and more objective bone age assessment.
Authors: Zhiheng Zhang, Jiajun Yang, Hong Sun, Dong Wang, Honghua Jiang, Yaru Chen, Tangyuan Ning
Abstract: Vegetation index (VI) saturation during the dense canopy stage and limited ground-truth annotations of winter wheat constrain accurate estimation of LAI and SPAD. Existing VI-based and texture-driven machine learning methods exhibit limited feature expressiveness. In addition, deep learning baselines suffer from domain gaps and high data demands, which restrict their generalization. Therefore, this study proposes the Multi-Channel Vegetation Indices Saturation Aware Net (MCVI-SANet), a lightweight semi-supervised vision model. The model incorporates a newly designed Vegetation Index Saturation-Aware Block (VI-SABlock) for adaptive channel-spatial feature enhancement. It also integrates a VICReg-based semi-supervised strategy to further improve generalization. Datasets were partitioned using a vegetation height-informed strategy to maintain representativeness across growth stages. Experiments over 10 repeated runs demonstrate that MCVI-SANet achieves state-of-the-art accuracy. The model attains an average R2 of 0.8123 and RMSE of 0.4796 for LAI, and an average R2 of 0.6846 and RMSE of 2.4222 for SPAD. This performance surpasses the best-performing baselines, with improvements of 8.95% in average LAI R2 and 8.17% in average SPAD R2. Moreover, MCVI-SANet maintains high inference speed with only 0.10M parameters. Overall, the integration of semi-supervised learning with agronomic priors provides a promising approach for enhancing remote sensing-based precision agriculture.
Authors: Dunxing Zhang (Technical University of Munich, Munich, Germany), Jiachen Lu (Technical University of Munich, Munich, Germany), Han Yang (National Science Center for Earthquake Engineering, Tianjin University, Tianjin, China, School of Civil Engineering, Tianjin University, Tianjin, China), Lei Bao (National Science Center for Earthquake Engineering, Tianjin University, Tianjin, China, School of Civil Engineering, Tianjin University, Tianjin, China), Bo Song (National Science Center for Earthquake Engineering, Tianjin University, Tianjin, China, School of Civil Engineering, Tianjin University, Tianjin, China)
Abstract: We propose ESSC-RM, a plug-and-play Enhancing framework for Semantic Scene Completion with a Refinement Module, which can be seamlessly integrated into existing SSC models. ESSC-RM operates in two phases: a baseline SSC network first produces a coarse voxel prediction, which is subsequently refined by a 3D U-Net-based Prediction Noise-Aware Module (PNAM) and Voxel-level Local Geometry Module (VLGM) under multiscale supervision. Experiments on SemanticKITTI show that ESSC-RM consistently improves semantic prediction performance. When integrated into CGFormer and MonoScene, the mean IoU increases from 16.87% to 17.27% and from 11.08% to 11.51%, respectively. These results demonstrate that ESSC-RM serves as a general refinement framework applicable to a wide range of SSC models.
Authors: Badr Moufad, Navid Bagheri Shouraki, Alain Oliviero Durmus, Thomas Hirtz, Eric Moulines, Jimmy Olsson, Yazid Janati
Abstract: Diffusion models have emerged as powerful priors for image editing tasks such as inpainting and local modification, where the objective is to generate realistic content that remains consistent with observed regions. In particular, zero-shot approaches that leverage a pretrained diffusion model, without any retraining, have been shown to achieve highly effective reconstructions. However, state-of-the-art zero-shot methods typically rely on a sequence of surrogate likelihood functions, whose scores are used as proxies for the ideal score. This procedure however requires vector-Jacobian products through the denoiser at every reverse step, introducing significant memory and runtime overhead. To address this issue, we propose a new likelihood surrogate that yields simple and efficient to sample Gaussian posterior transitions, sidestepping the backpropagation through the denoiser network. Our extensive experiments show that our method achieves strong observation consistency compared with fine-tuned baselines and produces coherent, high-quality reconstructions, all while significantly reducing inference cost. Code is available at https://github.com/YazidJanati/ding.
Authors: Wenhao Hu, Haonan Zhou, Zesheng Li, Liu Liu, Jiacheng Dong, Zhizhong Su, Gaoang Wang
Abstract: Recent advances in 3D scene representations have enabled high-fidelity novel view synthesis, yet adapting to discrete scene changes and constructing interactive 3D environments remain open challenges in vision and robotics. Existing approaches focus solely on updating a single scene without supporting novel-state synthesis. Others rely on diffusion-based object-background decoupling that works on one state at a time and cannot fuse information across multiple observations. To address these limitations, we introduce RecurGS, a recurrent fusion framework that incrementally integrates discrete Gaussian scene states into a single evolving representation capable of interaction. RecurGS detects object-level changes across consecutive states, aligns their geometric motion using semantic correspondence and Lie-algebra based SE(3) refinement, and performs recurrent updates that preserve historical structures through replay supervision. A voxelized, visibility-aware fusion module selectively incorporates newly observed regions while keeping stable areas fixed, mitigating catastrophic forgetting and enabling efficient long-horizon updates. RecurGS supports object-level manipulation, synthesizes novel scene states without requiring additional scans, and maintains photorealistic fidelity across evolving environments. Extensive experiments across synthetic and real-world datasets demonstrate that our framework delivers high-quality reconstructions with substantially improved update efficiency, providing a scalable step toward continuously interactive Gaussian worlds.
Authors: Charilaos Kapelonis, Marios Antonakakis, Konstantinos Politof, Aristomenis Antoniadis, Michalis Zervakis
Abstract: Art is widely recognized as a reflection of civilization and mosaics represent an important part of cultural heritage. Mosaics are an ancient art form created by arranging small pieces, called tesserae, on a surface using adhesive. Due to their age and fragility, they are prone to damage, highlighting the need for digital preservation. This paper addresses the problem of digitizing mosaics by segmenting the tesserae to separate them from the background within the broader field of Image Segmentation in Computer Vision. We propose a method leveraging Segment Anything Model 2 (SAM 2) by Meta AI, a foundation model that outperforms most conventional segmentation models, to automatically segment mosaics. Due to the limited open datasets in the field, we also create an annotated dataset of mosaic images to fine-tune and evaluate the model. Quantitative evaluation on our testing dataset shows notable improvements compared to the baseline SAM 2 model, with Intersection over Union increasing from 89.00% to 91.02% and Recall from 92.12% to 95.89%. Additionally, on a benchmark proposed by a prior approach, our model achieves an F-measure 3% higher than previous methods and reduces the error in the absolute difference between predicted and actual tesserae from 0.20 to just 0.02. The notable performance of the fine-tuned SAM 2 model together with the newly annotated dataset can pave the way for real-time segmentation of mosaic images.
Authors: Dimitrios Georgoulopoulos, Nikolaos Chaidos, Angeliki Dimitriou, Giorgos Stamou
Abstract: Accurately retrieving images that are semantically similar remains a fundamental challenge in computer vision, as traditional methods often fail to capture the relational and contextual nuances of a scene. We introduce PRISm (Pruning-based Image Retrieval via Importance Prediction on Semantic Graphs), a multimodal framework that advances image-to-image retrieval through two novel components. First, the Importance Prediction Module identifies and retains the most critical objects and relational triplets within an image while pruning irrelevant elements. Second, the Edge-Aware Graph Neural Network explicitly encodes relational structure and integrates global visual features to produce semantically informed image embeddings. PRISm achieves image retrieval that closely aligns with human perception by explicitly modeling the semantic importance of objects and their interactions, capabilities largely absent in prior approaches. Its architecture effectively combines relational reasoning with visual representation, enabling semantically grounded retrieval. Extensive experiments on benchmark and real-world datasets demonstrate consistently superior top-ranked performance, while qualitative analyses show that PRISm accurately captures key objects and interactions, producing interpretable and semantically meaningful results.
Authors: Fei Song, Yi Li, Jiangmeng Li, Rui Wang, Changwen Zheng, Fanjiang Xu, Hui Xiong
Abstract: Multi-prompt learning methods have emerged as an effective approach for facilitating the rapid adaptation of vision-language models to downstream tasks with limited resources. Existing multi-prompt learning methods primarily focus on utilizing various meticulously designed prompts within a single foundation vision-language model to achieve superior performance. However, the overlooked model-prompt matching bias hinders the development of multi-prompt learning, i.e., the same prompt can convey different semantics across distinct vision-language models, such as CLIP-ViT-B/16 and CLIP-ViT-B/32, resulting in inconsistent predictions of identical prompt. To mitigate the impact of this bias on downstream tasks, we explore an ensemble learning approach to sufficiently aggregate the benefits of diverse predictions. Additionally, we further disclose the presence of sample-prompt matching bias, which originates from the prompt-irrelevant semantics encapsulated in the input samples. Thus, directly utilizing all information from the input samples for generating weights of ensemble learning can lead to suboptimal performance. In response, we extract prompt-relevant semantics from input samples by leveraging the guidance of the information theory-based analysis, adaptively calculating debiased ensemble weights. Overall, we propose Adaptive-Debiased Ensemble MultiPrompt Learning, abbreviated as AmPLe, to mitigate the two types of bias simultaneously. Extensive experiments on three representative tasks, i.e., generalization to novel classes, new target datasets, and unseen domain shifts, show that AmPLe can widely outperform existing methods. Theoretical validation from a causal perspective further supports the effectiveness of AmPLe.
Authors: Seyed Ehsan Marjani Bajestani, Giovanni Beltrame
Abstract: Event-based cameras (ECs) have emerged as bio-inspired sensors that report pixel brightness changes asynchronously, offering unmatched speed and efficiency in vision sensing. Despite their high dynamic range, temporal resolution, low power consumption, and computational simplicity, traditional monochrome ECs face limitations in detecting static or slowly moving objects and lack color information essential for certain applications. To address these challenges, we present a novel approach that integrates a Digital Light Processing (DLP) projector, forming Active Structured Light (ASL) for RGB-D sensing. By combining the benefits of ECs and projection-based techniques, our method enables the detection of color and the depth of each pixel separately. Dynamic projection adjustments optimize bandwidth, ensuring selective color data acquisition and yielding colorful point clouds without sacrificing spatial resolution. This integration, facilitated by a commercial TI LightCrafter 4500 projector and a monocular monochrome EC, not only enables frameless RGB-D sensing applications but also achieves remarkable performance milestones. With our approach, we achieved a color detection speed equivalent to 1400 fps and 4 kHz of pixel depth detection, significantly advancing the realm of computer vision across diverse fields from robotics to 3D reconstruction methods. Our code is publicly available: https://github.com/MISTLab/event_based_rgbd_ros
Authors: Shurui Xu, Siqi Yang, Jiapin Ren, Zhong Cao, Hongwei Yang, Mengzhen Fan, Yuyu Sun, Shuyan Li
Abstract: Precise grading of meniscal horn tears is critical in knee injury diagnosis but remains underexplored in automated MRI analysis. Existing methods often rely on coarse study-level labels or binary classification, lacking localization and severity information. In this paper, we introduce MeniMV, a multi-view benchmark dataset specifically designed for horn-specific meniscus injury grading. MeniMV comprises 3,000 annotated knee MRI exams from 750 patients across three medical centers, providing 6,000 co-registered sagittal and coronal images. Each exam is meticulously annotated with four-tier (grade 0-3) severity labels for both anterior and posterior meniscal horns, verified by chief orthopedic physicians. Notably, MeniMV offers more than double the pathology-labeled data volume of prior datasets while uniquely capturing the dual-view diagnostic context essential in clinical practice. To demonstrate the utility of MeniMV, we benchmark multiple state-of-the-art CNN and Transformer-based models. Our extensive experiments establish strong baselines and highlight challenges in severity grading, providing a valuable foundation for future research in automated musculoskeletal imaging.
Authors: Zongyao Li, Yongkang Wong, Satoshi Yamazaki, Jianquan Liu, Mohan Kankanhalli
Abstract: Most existing video moment retrieval methods rely on temporal sequences of frame- or clip-level features that primarily encode global visual and semantic information. However, such representations often fail to capture fine-grained object semantics and appearance, which are crucial for localizing moments described by object-oriented queries involving specific entities and their interactions. In particular, temporal dynamics at the object level have been largely overlooked, limiting the effectiveness of existing approaches in scenarios requiring detailed object-level reasoning. To address this limitation, we propose a novel object-centric framework for moment retrieval. Our method first extracts query-relevant objects using a scene graph parser and then generates scene graphs from video frames to represent these objects and their relationships. Based on the scene graphs, we construct object-level feature sequences that encode rich visual and semantic information. These sequences are processed by a relational tracklet transformer, which models spatio-temporal correlations among objects over time. By explicitly capturing object-level state changes, our framework enables more accurate localization of moments aligned with object-oriented queries. We evaluated our method on three benchmarks: Charades-STA, QVHighlights, and TACoS. Experimental results demonstrate that our method outperforms existing state-of-the-art methods across all benchmarks.
Authors: Tianyang Zhanng, Xinxing Cheng, Jun Cheng, Shaoming Zheng, He Zhao, Huazhu Fu, Alejandro F Frangi, Jiang Liu, Jinming Duan
Abstract: Domain gaps arising from variations in imaging devices and population distributions pose significant challenges for machine learning in medical image analysis. Existing image-to-image translation methods primarily aim to learn mappings between domains, often generating diverse synthetic data with variations in anatomical scale and shape, but they usually overlook spatial correspondence during the translation process. For clinical applications, traceability, defined as the ability to provide pixel-level correspondences between original and translated images, is equally important. This property enhances clinical interpretability but has been largely overlooked in previous approaches. To address this gap, we propose Plasticine, which is, to the best of our knowledge, the first end-to-end image-to-image translation framework explicitly designed with traceability as a core objective. Our method combines intensity translation and spatial transformation within a denoising diffusion framework. This design enables the generation of synthetic images with interpretable intensity transitions and spatially coherent deformations, supporting pixel-wise traceability throughout the translation process.
Authors: Xiaoyang Guo, Keze Wang
Abstract: In recent years, large-scale vision-language models (VLMs) have demonstrated remarkable performance on multimodal understanding and reasoning tasks. However, handling high-dimensional visual features often incurs substantial computational and memory costs. VoCo-LLaMA alleviates this issue by compressing visual patch tokens into a few VoCo tokens, reducing computational overhead while preserving strong cross-modal alignment. Nevertheless, such approaches typically adopt a fixed compression rate, limiting their ability to adapt to varying levels of visual complexity. To address this limitation, we propose Adaptive-VoCo, a framework that augments VoCo-LLaMA with a lightweight predictor for adaptive compression. This predictor dynamically selects an optimal compression rate by quantifying an image's visual complexity using statistical cues from the vision encoder, such as patch token entropy and attention map variance. Furthermore, we introduce a joint loss function that integrates rate regularization with complexity alignment. This enables the model to balance inference efficiency with representational capacity, particularly in challenging scenarios. Experimental results show that our method consistently outperforms fixed-rate baselines across multiple multimodal tasks, highlighting the potential of adaptive visual compression for creating more efficient and robust VLMs.
Authors: Santwana Sagnika, Manav Malhotra, Ishtaj Kaur Deol, Soumyajit Roy, Swarnav Kumar
Abstract: Plant diseases pose a significant threat to agricultural productivity and global food security, accounting for 70-80% of crop losses worldwide. Traditional detection methods rely heavily on expert visual inspection, which is time-consuming, labour-intensive, and often impractical for large-scale farming operations. In this paper, we present PlantDiseaseNet-RT50, a novel fine-tuned deep learning architecture based on ResNet50 for automated plant disease detection. Our model features strategically unfrozen layers, a custom classification head with regularization mechanisms, and dynamic learning rate scheduling through cosine decay. Using a comprehensive dataset of distinct plant disease categories across multiple crop species, PlantDiseaseNet-RT50 achieves exceptional performance with approximately 98% accuracy, precision, and recall. Our architectural modifications and optimization protocol demonstrate how targeted fine-tuning can transform a standard pretrained model into a specialized agricultural diagnostic tool. We provide a detailed account of our methodology, including the systematic unfreezing of terminal layers, implementation of batch normalization and dropout regularization and application of advanced training techniques. PlantDiseaseNet-RT50 represents a significant advancement in AI-driven agricultural tools, offering a computationally efficient solution for rapid and accurate plant disease diagnosis that can be readily implemented in practical farming contexts to support timely interventions and reduce crop losses.
Authors: Benyamin Hosseiny, Kamirul Kamirul, Odysseas Pappas, Alin Achim
Abstract: Synthetic Aperture Radar (SAR) offers a unique capability for all-weather, space-based maritime activity monitoring by capturing and imaging strong reflections from ships at sea. A well-defined challenge in this domain is ship type classification. Due to the high diversity and complexity of ship types, accurate recognition is difficult and typically requires specialized deep learning models. These models, however, depend on large, high-quality ground-truth datasets to achieve robust performance and generalization. Furthermore, the growing variety of SAR satellites operating at different frequencies and spatial resolutions has amplified the need for more annotated datasets to enhance model accuracy. To address this, we present the NovaSAR Automated Ship Target Recognition (NASTaR) dataset. This dataset comprises of 3415 ship patches extracted from NovaSAR S-band imagery, with labels matched to AIS data. It includes distinctive features such as 23 unique classes, inshore/offshore separation, and an auxiliary wake dataset for patches where ship wakes are visible. We validated the dataset applicability across prominent ship-type classification scenarios using benchmark deep learning models. Results demonstrate over 60% accuracy for classifying four major ship types, over 70% for a three-class scenario, more than 75% for distinguishing cargo from tanker ships, and over 87% for identifying fishing vessels. The NASTaR dataset is available at https://10.5523/bris, while relevant codes for benchmarking and analysis are available at https://github.com/benyaminhosseiny/nastar.
URLs: https://10.5523/bris,, https://github.com/benyaminhosseiny/nastar.
Authors: Jensen Zhang, Ningyuan Liu, Keze Wang
Abstract: Vision-language models (VLMs) struggle in open-world applications, where out-of-distribution (OOD) concepts often trigger cross-modal alignment collapse and severely degrade zero-shot performance. We identify the root cause as modal asymmetry: while the visual encoder can extract discriminative features from unseen images, the text encoder is constrained by a fixed discrete vocabulary and cannot synthesize new semantic anchors. Existing approaches such as CoOp or LoRA provide only partial remedies, as they remain confined to the pre-trained semantic space. To overcome this bottleneck, we propose dynamic representation optimization, realized through the Guided Target-Matching Adaptation (GTMA) framework. At inference time, GTMA constructs a continuous pseudo-word embedding that best aligns with an OOD image's visual anchor, effectively bypassing vocabulary limitations. The optimization is driven by an adaptive gradient-based representation policy optimization algorithm, which incorporates semantic regularization to preserve plausibility and compatibility with the model's prior knowledge. Experiments on ImageNet-R and the VISTA-Beyond benchmark demonstrate that GTMA improves zero-shot and few-shot OOD accuracy by up to 15-20 percent over the base VLM while maintaining performance on in-distribution concepts. Ablation studies further confirm the necessity of pseudo-word optimization.
Authors: Rahul Yumlembam, Biju Issac, Nauman Aslam, Eaby Kollonoor Babu, Josh Collyer, Fraser Kennedy
Abstract: As AI-generated images become increasingly photorealistic, distinguishing them from natural images poses a growing challenge. This paper presents a robust detection framework that leverages multiple uncertainty measures to decide whether to trust or reject a model's predictions. We focus on three complementary techniques: Fisher Information, which captures the sensitivity of model parameters to input variations; entropy-based uncertainty from Monte Carlo Dropout, which reflects predictive variability; and predictive variance from a Deep Kernel Learning framework using a Gaussian Process classifier. To integrate these diverse uncertainty signals, Particle Swarm Optimisation is used to learn optimal weightings and determine an adaptive rejection threshold. The model is trained on Stable Diffusion-generated images and evaluated on GLIDE, VQDM, Midjourney, BigGAN, and StyleGAN3, each introducing significant distribution shifts. While standard metrics such as prediction probability and Fisher-based measures perform well in distribution, their effectiveness degrades under shift. In contrast, the Combined Uncertainty measure consistently achieves an incorrect rejection rate of approximately 70 percent on unseen generators, successfully filtering most misclassified AI samples. Although the system occasionally rejects correct predictions from newer generators, this conservative behaviour is acceptable, as rejected samples can support retraining. The framework maintains high acceptance of accurate predictions for natural images and in-domain AI data. Under adversarial attacks using FGSM and PGD, the Combined Uncertainty method rejects around 61 percent of successful attacks, while GP-based uncertainty alone achieves up to 80 percent. Overall, the results demonstrate that multi-source uncertainty fusion provides a resilient and adaptive solution for AI-generated image detection.
Authors: Moses Kiprono
Abstract: Chronic wounds, including diabetic foot ulcers which affect up to one-third of people with diabetes, impose a substantial clinical and economic burden, with U.S. healthcare costs exceeding 25 billion dollars annually. Current wound assessment remains predominantly subjective, leading to inconsistent classification and delayed interventions. We present WoundNet-Ensemble, an Internet of Medical Things system leveraging a novel ensemble of three complementary deep learning architectures: ResNet-50, the self-supervised Vision Transformer DINOv2, and Swin Transformer, for automated classification of six clinically distinct wound types. Our system achieves 99.90 percent ensemble accuracy on a comprehensive dataset of 5,175 wound images spanning diabetic foot ulcers, pressure ulcers, venous ulcers, thermal burns, pilonidal sinus wounds, and fungating malignant tumors. The weighted fusion strategy demonstrates a 3.7 percent improvement over previous state-of-the-art methods. Furthermore, we implement a longitudinal wound healing tracker that computes healing rates, severity scores, and generates clinical alerts. This work demonstrates a robust, accurate, and clinically deployable tool for modernizing wound care through artificial intelligence, addressing critical needs in telemedicine and remote patient monitoring. The implementation and trained models will be made publicly available to support reproducibility.
Authors: Alexander M. Glandon, Khan M. Iftekharuddin
Abstract: Multisource domain adaptation (MDA) aims to use multiple source datasets with available labels to infer labels on a target dataset without available labels for target supervision. Prior works on MDA in the literature is ad-hoc as the pretraining of source models is either based on weight sharing or uses independently trained models. This work proposes a Bayesian framework for pretraining in MDA by considering that the distributions of different source domains are typically similar. The Hierarchical Bayesian Framework uses similarity between the different source data distributions to optimize the pretraining for MDA. Experiments using the proposed Bayesian framework for MDA show that our framework improves accuracy on recognition tasks for a large benchmark dataset. Performance comparison with state-of-the-art MDA methods on the challenging problem of human action recognition in multi-domain benchmark Daily-DA RGB video shows the proposed Bayesian Framework offers a 17.29% improvement in accuracy when compared to the state-of-the-art methods in the literature.
Authors: Aofei Chang, Ting Wang, Fenglong Ma
Abstract: Medical Large Vision-Language Models (Med-LVLMs) have shown promising results in clinical applications, but often suffer from hallucinated outputs due to misaligned visual understanding. In this work, we identify two fundamental limitations contributing to this issue: insufficient visual representation learning and poor visual attention alignment. To address these problems, we propose MEDALIGN, a simple, lightweight alignment distillation framework that transfers visual alignment knowledge from a domain-specific Contrastive Language-Image Pre-training (CLIP) model to Med-LVLMs. MEDALIGN introduces two distillation losses: a spatial-aware visual alignment loss based on visual token-level similarity structures, and an attention-aware distillation loss that guides attention toward diagnostically relevant regions. Extensive experiments on medical report generation and medical visual question answering (VQA) benchmarks show that MEDALIGN consistently improves both performance and interpretability, yielding more visually grounded outputs.
Authors: Qixiang Chen, Cheng Zhang, Chi-Wing Fu, Jingwen Ye, Jianfei Cai
Abstract: Recent multimodal large language models (MLLMs) show great potential in natural image understanding. Yet, they perform well, mainly on reasoning in-view contents within the image frame. This paper presents the first study on out-of-view (OOV) understanding, i.e., the ability to reason objects, activities, and scenes beyond the visible frame of a perspective view. Our technical contributions are threefold. First, we design OpenView, a four-stage pipeline to massively generate multi-choice VQA by leveraging panoramic imagery to enable context-rich and spatial-grounded VQA synthesis with free-view framing. Second, we curate OpenView-Dataset, a high-quality synthetic dataset from diverse real-world panoramas to empower MLLMs upon supervised fine-tuning. Third, we build OpenView-Bench, a benchmark that jointly measures choice and rationale accuracy for interpretable and diagnosable evaluation. Experimental results show that despite having a large gap from human performance in OOV VQA answer selection, upon empowered by OpenView, multiple MLLMs can consistently boost their performance, uplifted from 48.6% to 64.1% on average. Code, benchmark, and data will be available at https://github.com/q1xiangchen/OpenView.
Authors: Sumaiya Ali, Areej Alhothali, Ohoud Alzamzami, Sameera Albasri, Ahmed Abduljabbar, Muhammad Alwazzan
Abstract: Placenta Accreta Spectrum (PAS) is a serious obstetric condition that can be challenging to diagnose with Magnetic Resonance Imaging (MRI) due to variability in radiologists' interpretations. To overcome this challenge, a hybrid 3D deep learning model for automated PAS detection from volumetric MRI scans is proposed in this study. The model integrates a 3D DenseNet121 to capture local features and a 3D Vision Transformer (ViT) to model global spatial context. It was developed and evaluated on a retrospective dataset of 1,133 MRI volumes. Multiple 3D deep learning architectures were also evaluated for comparison. On an independent test set, the DenseNet121-ViT model achieved the highest performance with a five-run average accuracy of 84.3%. These results highlight the strength of hybrid CNN-Transformer models as a computer-aided diagnosis tool. The model's performance demonstrates a clear potential to assist radiologists by providing a robust decision support to improve diagnostic consistency across interpretations, and ultimately enhance the accuracy and timeliness of PAS diagnosis.
Authors: Zhe Li, Kun Cheng, Hanyue Mo, Jintao Lu, Ziwen Kuang, Jianwen Ye, Lixu Xu, Xinya Meng, Jiahui Zhao, Shengda Ji, Shuyuan Liu, Mengyu Wang
Abstract: A vision-based trajectory analysis solution is proposed to address the "zero-speed braking" issue caused by inaccurate Controller Area Network (CAN) signals in commercial vehicle Automatic Emergency Braking (AEB) systems during low-speed operation. The algorithm utilizes the NVIDIA Jetson AGX Xavier platform to process sequential video frames from a blind spot camera, employing self-adaptive Contrast Limited Adaptive Histogram Equalization (CLAHE)-enhanced Scale-Invariant Feature Transform (SIFT) feature extraction and K-Nearest Neighbors (KNN)-Random Sample Consensus (RANSAC) matching. This allows for precise classification of the vehicle's motion state (static, vibration, moving). Key innovations include 1) multiframe trajectory displacement statistics (5-frame sliding window), 2) a dual-threshold state decision matrix, and 3) OBD-II driven dynamic Region of Interest (ROI) configuration. The system effectively suppresses environmental interference and false detection of dynamic objects, directly addressing the challenge of low-speed false activation in commercial vehicle safety systems. Evaluation in a real-world dataset (32,454 video segments from 1,852 vehicles) demonstrates an F1-score of 99.96% for static detection, 97.78% for moving state recognition, and a processing delay of 14.2 milliseconds (resolution 704x576). The deployment on-site shows an 89% reduction in false braking events, a 100% success rate in emergency braking, and a fault rate below 5%.
Authors: Jianglin Lu, Yuanwei Wu, Ziyi Zhao, Hongcheng Wang, Felix Jimenez, Abrar Majeedi, Yun Fu
Abstract: Complex image restoration aims to recover high-quality images from inputs affected by multiple degradations such as blur, noise, rain, and compression artifacts. Recent restoration agents, powered by vision-language models and large language models, offer promising restoration capabilities but suffer from significant efficiency bottlenecks due to reflection, rollback, and iterative tool searching. Moreover, their performance heavily depends on degradation recognition models that require extensive annotations for training, limiting their applicability in label-free environments. To address these limitations, we propose a policy optimization-based restoration framework that learns an lightweight agent to determine tool-calling sequences. The agent operates in a sequential decision process, selecting the most appropriate restoration operation at each step to maximize final image quality. To enable training within label-free environments, we introduce a novel reward mechanism driven by multimodal large language models, which act as human-aligned evaluator and provide perceptual feedback for policy improvement. Once trained, our agent executes a deterministic restoration plans without redundant tool invocations, significantly accelerating inference while maintaining high restoration quality. Extensive experiments show that despite using no supervision, our method matches SOTA performance on full-reference metrics and surpasses existing approaches on no-reference metrics across diverse degradation scenarios.
Authors: Saeideh Yousefzadeh, Hamidreza Pourreza
Abstract: Visual Place Recognition (VPR) in long-term deployment requires reasoning beyond pixel similarity: systems must make transparent, interpretable decisions that remain robust under lighting, weather and seasonal change. We present Text2Graph VPR, an explainable semantic localization system that converts image sequences into textual scene descriptions, parses those descriptions into structured scene graphs, and reasons over the resulting graphs to identify places. Scene graphs capture objects, attributes and pairwise relations; we aggregate per-frame graphs into a compact place representation and perform retrieval with a dual-similarity mechanism that fuses learned Graph Attention Network (GAT) embeddings and a Shortest-Path (SP) kernel for structural matching. This hybrid design enables both learned semantic matching and topology-aware comparison, and -- critically -- produces human-readable intermediate representations that support diagnostic analysis and improve transparency in the decision process. We validate the system on Oxford RobotCar and MSLS (Amman/San Francisco) benchmarks and demonstrate robust retrieval under severe appearance shifts, along with zero-shot operation using human textual queries. The results illustrate that semantic, graph-based reasoning is a viable and interpretable alternative for place recognition, particularly suited to safety-sensitive and resource-constrained settings.
Authors: Ruiqi Chen, Kaitong Cai, Yijia Fan, Keze Wang
Abstract: Traditional animation production involves complex pipelines and significant manual labor cost. While recent video generation models such as Sora, Kling, and CogVideoX achieve impressive results on natural video synthesis, they exhibit notable limitations when applied to animation generation. Recent efforts, such as AniSora, demonstrate promising performance by fine-tuning image-to-video models for animation styles, yet analogous exploration in the text-to-video setting remains limited. In this work, we present PTTA, a pure text-to-animation framework for high-quality animation creation. We first construct a small-scale but high-quality paired dataset of animation videos and textual descriptions. Building upon the pretrained text-to-video model HunyuanVideo, we perform fine-tuning to adapt it to animation-style generation. Extensive visual evaluations across multiple dimensions show that the proposed approach consistently outperforms comparable baselines in animation video synthesis.
Authors: Xiyue Bai, Ronghao Yu, Jia Xiu, Pengfei Zhou, Jie Xia, Peng Ji
Abstract: Generating or editing images directly from Neural signals has immense potential at the intersection of neuroscience, vision, and Brain-computer interaction. In this paper, We present Uni-Neur2Img, a unified framework for neural signal-driven image generation and editing. The framework introduces a parameter-efficient LoRA-based neural signal injection module that independently processes each conditioning signal as a pluggable component, facilitating flexible multi-modal conditioning without altering base model parameters. Additionally, we employ a causal attention mechanism accommodate the long-sequence modeling demands of conditional generation tasks. Existing neural-driven generation research predominantly focuses on textual modalities as conditions or intermediate representations, resulting in limited exploration of visual modalities as direct conditioning signals. To bridge this research gap, we introduce the EEG-Style dataset. We conduct comprehensive evaluations across public benchmarks and self-collected neural signal datasets: (1) EEG-driven image generation on the public CVPR40 dataset; (2) neural signal-guided image editing on the public Loongx dataset for semantic-aware local modifications; and (3) EEG-driven style transfer on our self-collected EEG-Style dataset. Extensive experimental results demonstrate significant improvements in generation fidelity, editing consistency, and style transfer quality while maintaining low computational overhead and strong scalability to additional modalities. Thus, Uni-Neur2Img offers a unified, efficient, and extensible solution for bridging neural signals and visual content generation.
Authors: Kai Kohyama, Yoshimitsu Aoki, Guillermo Gallego, Shintaro Shiba
Abstract: Event cameras offer a high temporal resolution over traditional frame-based cameras, which makes them suitable for motion and structure estimation. However, it has been unclear how event-based 3D Gaussian Splatting (3DGS) approaches could leverage fine-grained temporal information of sparse events. This work proposes a framework to address the trade-off between accuracy and temporal resolution in event-based 3DGS. Our key idea is to decouple the rendering into two branches: event-by-event geometry (depth) rendering and snapshot-based radiance (intensity) rendering, by using ray-tracing and the image of warped events. The extensive evaluation shows that our method achieves state-of-the-art performance on the real-world datasets and competitive performance on the synthetic dataset. Also, the proposed method works without prior information (e.g., pretrained image reconstruction models) or COLMAP-based initialization, is more flexible in the event selection number, and achieves sharp reconstruction on scene edges with fast training time. We hope that this work deepens our understanding of the sparse nature of events for 3D reconstruction. The code will be released.
Authors: Zhiyuan Peng, Zihan Ye, Shreyank N Gowda, Yuping Yan, Haotian Xu, Ling Shao
Abstract: Zero-shot Learning (ZSL) aims to enable image classifiers to recognize images from unseen classes that were not included during training. Unlike traditional supervised classification, ZSL typically relies on learning a mapping from visual features to predefined, human-understandable class concepts. While ZSL models promise to improve generalization and interpretability, their robustness under systematic input perturbations remain unclear. In this study, we present an empirical analysis about the robustness of existing ZSL methods at both classlevel and concept-level. Specifically, we successfully disrupted their class prediction by the well-known non-target class attack (clsA). However, in the Generalized Zero-shot Learning (GZSL) setting, we observe that the success of clsA is only at the original best-calibrated point. After the attack, the optimal bestcalibration point shifts, and ZSL models maintain relatively strong performance at other calibration points, indicating that clsA results in a spurious attack success in the GZSL. To address this, we propose the Class-Bias Enhanced Attack (CBEA), which completely eliminates GZSL accuracy across all calibrated points by enhancing the gap between seen and unseen class probabilities.Next, at concept-level attack, we introduce two novel attack modes: Class-Preserving Concept Attack (CPconA) and NonClass-Preserving Concept Attack (NCPconA). Our extensive experiments evaluate three typical ZSL models across various architectures from the past three years and reveal that ZSL models are vulnerable not only to the traditional class attack but also to concept-based attacks. These attacks allow malicious actors to easily manipulate class predictions by erasing or introducing concepts. Our findings highlight a significant performance gap between existing approaches, emphasizing the need for improved adversarial robustness in current ZSL models.
Authors: Yue Wen, Liang Song, Hesheng Wang
Abstract: Low-light 3D reconstruction from sparse views remains challenging due to exposure imbalance and degraded color fidelity. While existing methods struggle with view inconsistency and require per-scene training, we propose SplatBright, which is, to our knowledge, the first generalizable 3D Gaussian framework for joint low-light enhancement and reconstruction from sparse sRGB inputs. Our key idea is to integrate physically guided illumination modeling with geometry-appearance decoupling for consistent low-light reconstruction. Specifically, we adopt a dual-branch predictor that provides stable geometric initialization of 3D Gaussian parameters. On the appearance side, illumination consistency leverages frequency priors to enable controllable and cross-view coherent lighting, while an appearance refinement module further separates illumination, material, and view-dependent cues to recover fine texture. To tackle the lack of large-scale geometrically consistent paired data, we synthesize dark views via a physics-based camera model for training. Extensive experiments on public and self-collected datasets demonstrate that SplatBright achieves superior novel view synthesis, cross-view consistency, and better generalization to unseen low-light scenes compared with both 2D and 3D methods.
Authors: Pengxiang Ouyang, Qing Ma, Zheng Wang, Cong Bai
Abstract: Remote sensing (RS) image-text retrieval faces significant challenges in real-world datasets due to the presence of Pseudo-Matched Pairs (PMPs), semantically mismatched or weakly aligned image-text pairs, which hinder the learning of reliable cross-modal alignments. To address this issue, we propose a novel retrieval framework that leverages Cross-Modal Gated Attention and a Positive-Negative Awareness Attention mechanism to mitigate the impact of such noisy associations. The gated module dynamically regulates cross-modal information flow, while the awareness mechanism explicitly distinguishes informative (positive) cues from misleading (negative) ones during alignment learning. Extensive experiments on three benchmark RS datasets, i.e., RSICD, RSITMD, and RS5M, demonstrate that our method consistently achieves state-of-the-art performance, highlighting its robustness and effectiveness in handling real-world mismatches and PMPs in RS image-text retrieval tasks.
Authors: Yiming Sun, Mi Zhang, Feifei Li, Geng Hong, Min Yang
Abstract: Despite Video Large Language Models having rapidly advanced in recent years, perceptual hallucinations pose a substantial safety risk, which severely restricts their real-world applicability. While several methods for hallucination mitigation have been proposed, they often compromise the model's capacity for video understanding and reasoning. In this work, we propose SmartSight, a pioneering step to address this issue in a training-free manner by leveraging the model's own introspective capabilities. Specifically, SmartSight generates multiple candidate responses to uncover low-hallucinated outputs that are often obscured by standard greedy decoding. It assesses the hallucination of each response using the Temporal Attention Collapse score, which measures whether the model over-focuses on trivial temporal regions of the input video when generating the response. To improve efficiency, SmartSight identifies the Visual Attention Vanishing point, enabling more accurate hallucination estimation and early termination of hallucinated responses, leading to a substantial reduction in decoding cost. Experiments show that SmartSight substantially lowers hallucinations for Qwen2.5-VL-7B by 10.59% on VRIPT-HAL, while simultaneously enhancing video understanding and reasoning, boosting performance on VideoMMMU by up to 8.86%. These results highlight SmartSight's effectiveness in improving the reliability of open-source Video-LLMs.
Authors: Longhuan Xu, Feng Yin, Cunjian Chen
Abstract: Text-to-image diffusion inference typically follows synchronized schedules, where the numerical integrator advances the latent state to the same timestep at which the denoiser is conditioned. We propose an asynchronous inference mechanism that decouples these two, allowing the denoiser to be conditioned at a different, learned timestep while keeping image update schedule unchanged. A lightweight timestep prediction module (TPM), trained with Group Relative Policy Optimization (GRPO), selects a more feasible conditioning timestep based on the current state, effectively choosing a desired noise level to control image detail and textural richness. At deployment, a scaling hyper-parameter can be used to interpolate between the original and de-synchronized timesteps, enabling conservative or aggressive adjustments. To keep the study computationally affordable, we cap the inference at 15 steps for SD3.5 and 10 steps for Flux. Evaluated on Stable Diffusion 3.5 Medium and Flux.1-dev across MS-COCO 2014 and T2I-CompBench datasets, our method optimizes a composite reward that averages Image Reward, HPSv2, CLIP Score and Pick Score, and shows consistent improvement.
Authors: Maxime Kayser, Maksim Gridnev, Wanting Wang, Max Bain, Aneesh Rangnekar, Avijit Chatterjee, Aleksandr Petrov, Harini Veeraraghavan, Nathaniel C. Swinburne
Abstract: We present brat (brain report alignment transformer), a multi-view representation learning framework for brain magnetic resonance imaging (MRI) trained on MRIs paired with clinical reports. Brain MRIs present unique challenges due to the presence of numerous, highly varied, and often subtle abnormalities that are localized to a few slices within a 3D volume. To address these challenges, we introduce a brain MRI dataset $10\times$ larger than existing ones, containing approximately 80,000 3D scans with corresponding radiology reports, and propose a multi-view pre-training approach inspired by advances in document retrieval. We develop an implicit query-feature matching mechanism and adopt concepts from quality-diversity to obtain multi-view embeddings of MRIs that are aligned with the clinical features given by report sentences. We evaluate our approach across multiple vision-language and vision tasks, demonstrating substantial performance improvements. The brat foundation models are publicly released.
Authors: Huimin Wu, Kwang-Ting Cheng, Stephen Lin, Zhirong Wu
Abstract: This paper presents an investigation of vision transformer learning for multi-view geometry tasks, such as optical flow estimation, by fine-tuning video foundation models. Unlike previous methods that involve custom architectural designs and task-specific pretraining, our research finds that general-purpose models pretrained on videos can be readily transferred to multi-view problems with minimal adaptation. The core insight is that general-purpose attention between patches learns temporal and spatial information for geometric reasoning. We demonstrate that appending a linear decoder to the Transformer backbone produces satisfactory results, and iterative refinement can further elevate performance to stateof-the-art levels. This conceptually simple approach achieves top cross-dataset generalization results for optical flow estimation with end-point error (EPE) of 0.69, 1.78, and 3.15 on the Sintel clean, Sintel final, and KITTI datasets, respectively. Our method additionally establishes a new record on the online test benchmark with EPE values of 0.79, 1.88, and F1 value of 3.79. Applications to 3D depth estimation and stereo matching also show strong performance, illustrating the versatility of video-pretrained models in addressing geometric vision tasks.
Authors: Jongmin Park, Minh-Quan Viet Bui, Juan Luis Gonzalez Bello, Jaeho Moon, Jihyong Oh, Munchurl Kim
Abstract: Feed-forward 3D Gaussian Splatting (3DGS) enables efficient one-pass scene reconstruction, providing 3D representations for novel view synthesis without per-scene optimization. However, existing methods typically predict pixel-aligned primitives per-view, producing an excessive number of primitives in dense-view settings and offering no explicit control over the number of predicted Gaussians. To address this, we propose EcoSplat, the first efficiency-controllable feed-forward 3DGS framework that adaptively predicts the 3D representation for any given target primitive count at inference time. EcoSplat adopts a two-stage optimization process. The first stage is Pixel-aligned Gaussian Training (PGT) where our model learns initial primitive prediction. The second stage is Importance-aware Gaussian Finetuning (IGF) stage where our model learns rank primitives and adaptively adjust their parameters based on the target primitive count. Extensive experiments across multiple dense-view settings show that EcoSplat is robust and outperforms state-of-the-art methods under strict primitive-count constraints, making it well-suited for flexible downstream rendering tasks.
Authors: Linwei Qiu, Gongzhe Li, Xiaozhe Zhang, Qinlin Sun, Fengying Xie
Abstract: Image correction and rectangling are valuable tasks in practical photography systems such as smartphones. Recent remarkable advancements in deep learning have undeniably brought about substantial performance improvements in these fields. Nevertheless, existing methods mainly rely on task-specific architectures. This significantly restricts their generalization ability and effective application across a wide range of different tasks. In this paper, we introduce the Unified Rectification Framework (UniRect), a comprehensive approach that addresses these practical tasks from a consistent distortion rectification perspective. Our approach incorporates various task-specific inverse problems into a general distortion model by simulating different types of lenses. To handle diverse distortions, UniRect adopts one task-agnostic rectification framework with a dual-component structure: a {Deformation Module}, which utilizes a novel Residual Progressive Thin-Plate Spline (RP-TPS) model to address complex geometric deformations, and a subsequent Restoration Module, which employs Residual Mamba Blocks (RMBs) to counteract the degradation caused by the deformation process and enhance the fidelity of the output image. Moreover, a Sparse Mixture-of-Experts (SMoEs) structure is designed to circumvent heavy task competition in multi-task learning due to varying distortions. Extensive experiments demonstrate that our models have achieved state-of-the-art performance compared with other up-to-date methods.
Authors: Jinqiu Chen, Huyan Xu
Abstract: Predicting breast cancer recurrence risk is a critical clinical challenge. This study investigates the potential of computational pathology to stratify patients using deep learning on routine Hematoxylin and Eosin (H&E) stained whole-slide images (WSIs). We developed and compared three Multiple Instance Learning (MIL) frameworks -- CLAM-SB, ABMIL, and ConvNeXt-MIL-XGBoost -- on an in-house dataset of 210 patient cases. The models were trained to predict 5-year recurrence risk, categorized into three tiers (low, medium, high), with ground truth labels established by the 21-gene Recurrence Score. Features were extracted using the UNI and CONCH pre-trained models. In a 5-fold cross-validation, the modified CLAM-SB model demonstrated the strongest performance, achieving a mean Area Under the Curve (AUC) of 0.836 and a classification accuracy of 76.2%. Our findings demonstrate the feasibility of using deep learning on standard histology slides for automated, genomics-correlated risk stratification, highlighting a promising pathway toward rapid and cost-effective clinical decision support.
Authors: Kewei Wei, Bocheng Hu, Jie Cao, Xiaohan Chen, Zhengxi Lu, Wubing Xia, Weili Xu, Jiaao Wu, Junchen He, Mingyu Jia, Ciyun Zhao, Ye Sun, Yizhi Li, Zhonghan Zhao, Jian Zhang, Gaoang Wang
Abstract: Modern Large Multimodal Models (LMMs) have demonstrated extraordinary ability in static image and single-state spatial-temporal understanding. However, their capacity to comprehend the dynamic changes of objects within a shared spatial context between two distinct video observations, remains largely unexplored. This ability to reason about transformations within a consistent environment is particularly crucial for advancements in the field of spatial intelligence. In this paper, we introduce $M^3-Verse$, a Multi-Modal, Multi-State, Multi-Dimensional benchmark, to formally evaluate this capability. It is built upon paired videos that provide multi-perspective observations of an indoor scene before and after a state change. The benchmark contains a total of 270 scenes and 2,932 questions, which are categorized into over 50 subtasks that probe 4 core capabilities. We evaluate 16 state-of-the-art LMMs and observe their limitations in tracking state transitions. To address these challenges, we further propose a simple yet effective baseline that achieves significant performance improvements in multi-state perception. $M^3-Verse$ thus provides a challenging new testbed to catalyze the development of next-generation models with a more holistic understanding of our dynamic visual world. You can get the construction pipeline from https://github.com/Wal-K-aWay/M3-Verse_pipeline and full benchmark data from https://www.modelscope.cn/datasets/WalKaWay/M3-Verse.
URLs: https://github.com/Wal-K-aWay/M3-Verse_pipeline, https://www.modelscope.cn/datasets/WalKaWay/M3-Verse.
Authors: James E. Gallagher, Edward J. Oughton
Abstract: Landmines remain a persistent humanitarian threat, with an estimated 110 million mines deployed across 60 countries, claiming approximately 26,000 casualties annually. Current detection methods are hazardous, inefficient, and prohibitively expensive. We present the Adaptive Multispectral Landmine Identification Dataset (AMLID), the first open-source dataset combining Red-Green-Blue (RGB) and Long-Wave Infrared (LWIR) imagery for Unmanned Aerial Systems (UAS)-based landmine detection. AMLID comprises of 12,078 labeled images featuring 21 globally deployed landmine types across anti-personnel and anti-tank categories in both metal and plastic compositions. The dataset spans 11 RGB-LWIR fusion levels, four sensor altitudes, two seasonal periods, and three daily illumination conditions. By providing comprehensive multispectral coverage across diverse environmental variables, AMLID enables researchers to develop and benchmark adaptive detection algorithms without requiring access to live ordnance or expensive data collection infrastructure, thereby democratizing humanitarian demining research.
Authors: Tianrui Zhu, Shiyi Zhang, Zhirui Sun, Jingqi Tian, Yansong Tang
Abstract: Frame-level autoregressive (frame-AR) models have achieved significant progress, enabling real-time video generation comparable to bidirectional diffusion models and serving as a foundation for interactive world models and game engines. However, current approaches in long video generation typically rely on window attention, which naively discards historical context outside the window, leading to catastrophic forgetting and scene inconsistency; conversely, retaining full history incurs prohibitive memory costs. To address this trade-off, we propose \textbf{Memorize-and-Generate (MAG)}, a framework that decouples memory compression and frame generation into distinct tasks. Specifically, we train a memory model to compress historical information into a compact KV cache, and a separate generator model to synthesize subsequent frames utilizing this compressed representation. Furthermore, we introduce \textbf{MAG-Bench} to strictly evaluate historical memory retention. Extensive experiments demonstrate that MAG achieves superior historical scene consistency while maintaining competitive performance on standard video generation benchmarks.
Authors: Kaican Li, Lewei Yao, Jiannan Wu, Tiezheng Yu, Jierun Chen, Haoli Bai, Lu Hou, Lanqing Hong, Wei Zhang, Nevin L. Zhang
Abstract: The ability for AI agents to "think with images" requires a sophisticated blend of reasoning and perception. However, current open multimodal agents still largely fall short on the reasoning aspect crucial for real-world tasks like analyzing documents with dense charts/diagrams and navigating maps. To address this gap, we introduce O3-Bench, a new benchmark designed to evaluate multimodal reasoning with interleaved attention to visual details. O3-Bench features challenging problems that require agents to piece together subtle visual information from distinct image areas through multi-step reasoning. The problems are highly challenging even for frontier systems like OpenAI o3, which only obtains 40.8% accuracy on O3-Bench. To make progress, we propose InSight-o3, a multi-agent framework consisting of a visual reasoning agent (vReasoner) and a visual search agent (vSearcher) for which we introduce the task of generalized visual search -- locating relational, fuzzy, or conceptual regions described in free-form language, beyond just simple objects or figures in natural images. We then present a multimodal LLM purpose-trained for this task via reinforcement learning. As a plug-and-play agent, our vSearcher empowers frontier multimodal models (as vReasoners), significantly improving their performance on a wide range of benchmarks. This marks a concrete step towards powerful o3-like open systems. Our code and dataset can be found at https://github.com/m-Just/InSight-o3 .
Authors: Yuan Chen, Zichen Wen, Yuzhou Wu, Xuyang Liu, Shuang Chen, Junpeng Ma, Weijia Li, Conghui He, Linfeng Zhang
Abstract: Multimodal Large Language Models (MLLMs) deliver strong vision-language performance but at high computational cost, driven by numerous visual tokens processed by the Vision Transformer (ViT) encoder. Existing token pruning strategies are inadequate: LLM-stage token pruning overlooks the ViT's overhead, while conventional ViT token pruning, without language guidance, risks discarding textually critical visual cues and introduces feature distortions amplified by the ViT's bidirectional attention. To meet these challenges, we propose IPCV, a training-free, information-preserving compression framework for MLLM visual encoders. IPCV enables aggressive token pruning inside the ViT via Neighbor-Guided Reconstruction (NGR) that temporarily reconstructs pruned tokens to participate in attention with minimal overhead, then fully restores them before passing to the LLM. Besides, we introduce Attention Stabilization (AS) to further alleviate the negative influence from token pruning by approximating the K/V of pruned tokens. It can be directly applied to previous LLM-side token pruning methods to enhance their performance. Extensive experiments show that IPCV substantially reduces end-to-end computation and outperforms state-of-the-art training-free token compression methods across diverse image and video benchmarks. Our code is available at https://github.com/Perkzi/IPCV.
Authors: Xiaoyang Li, Wenzhu Yang, Kanglin Wang, Tiebiao Wang, Qingsong Fei
Abstract: Action recognition is a critical task in video understanding, requiring the comprehensive capture of spatio-temporal cues across various scales. However, existing methods often overlook the multi-granularity nature of actions. To address this limitation, we introduce the Context-Aware Network (CAN). CAN consists of two core modules: the Multi-scale Temporal Cue Module (MTCM) and the Group Spatial Cue Module (GSCM). MTCM effectively extracts temporal cues at multiple scales, capturing both fast-changing motion details and overall action flow. GSCM, on the other hand, extracts spatial cues at different scales by grouping feature maps and applying specialized extraction methods to each group. Experiments conducted on five benchmark datasets (Something-Something V1 and V2, Diving48, Kinetics-400, and UCF101) demonstrate the effectiveness of CAN. Our approach achieves competitive performance, outperforming most mainstream methods, with accuracies of 50.4% on Something-Something V1, 63.9% on Something-Something V2, 88.4% on Diving48, 74.9% on Kinetics-400, and 86.9% on UCF101. These results highlight the importance of capturing multi-scale spatio-temporal cues for robust action recognition.
Authors: Guohui Zhang, Hu Yu, Xiaoxiao Ma, Yaning Pan, Hang Xu, Feng Zhao
Abstract: Reinforcement learning (RL) has demonstrated significant potential for post-training language models and autoregressive visual generative models, but adapting RL to masked generative models remains challenging. The core factor is that policy optimization requires accounting for the probability likelihood of each step due to its multi-step and iterative refinement process. This reliance on entire sampling trajectories introduces high computational cost, whereas natively optimizing random steps often yields suboptimal results. In this paper, we present MaskFocus, a novel RL framework that achieves effective policy optimization for masked generative models by focusing on critical steps. Specifically, we determine the step-level information gain by measuring the similarity between the intermediate images at each sampling step and the final generated image. Crucially, we leverage this to identify the most critical and valuable steps and execute focused policy optimization on them. Furthermore, we design a dynamic routing sampling mechanism based on entropy to encourage the model to explore more valuable masking strategies for samples with low entropy. Extensive experiments on multiple Text-to-Image benchmarks validate the effectiveness of our method.
Authors: Wenze Liu, Weicai Ye, Minghong Cai, Quande Liu, Xintao Wang, Xiangyu Yue
Abstract: Recent advancements in video generation have seen a shift towards unified, transformer-based foundation models that can handle multiple conditional inputs in-context. However, these models have primarily focused on modalities like text, images, and depth maps, while strictly time-synchronous signals like audio have been underexplored. This paper introduces In-Context Audio Control of video diffusion transformers (ICAC), a framework that investigates the integration of audio signals for speech-driven video generation within a unified full-attention architecture, akin to FullDiT. We systematically explore three distinct mechanisms for injecting audio conditions: standard cross-attention, 2D self-attention, and unified 3D self-attention. Our findings reveal that while 3D attention offers the highest potential for capturing spatio-temporal audio-visual correlations, it presents significant training challenges. To overcome this, we propose a Masked 3D Attention mechanism that constrains the attention pattern to enforce temporal alignment, enabling stable training and superior performance. Our experiments demonstrate that this approach achieves strong lip synchronization and video quality, conditioned on an audio stream and reference images.
Authors: Fanis Mathioulakis, Gorjan Radevski, Tinne Tuytelaars
Abstract: We introduce Eff-GRot, an approach for efficient and generalizable rotation estimation from RGB images. Given a query image and a set of reference images with known orientations, our method directly predicts the object's rotation in a single forward pass, without requiring object- or category-specific training. At the core of our framework is a transformer that performs a comparison in the latent space, jointly processing rotation-aware representations from multiple references alongside a query. This design enables a favorable balance between accuracy and computational efficiency while remaining simple, scalable, and fully end-to-end. Experimental results show that Eff-GRot offers a promising direction toward more efficient rotation estimation, particularly in latency-sensitive applications.
Authors: Guangtao Lyu, Chenghao Xu, Qi Liu, Jiexi Yan, Muli Yang, Fen Fang, Cheng Deng
Abstract: Music to 3D dance generation aims to synthesize realistic and rhythmically synchronized human dance from music. While existing methods often rely on additional genre labels to further improve dance generation, such labels are typically noisy, coarse, unavailable, or insufficient to capture the diversity of real-world music, which can result in rhythm misalignment or stylistic drift. In contrast, we observe that tempo, a core property reflecting musical rhythm and pace, remains relatively consistent across datasets and genres, typically ranging from 60 to 200 BPM. Based on this finding, we propose TempoMoE, a hierarchical tempo-aware Mixture-of-Experts module that enhances the diffusion model and its rhythm perception. TempoMoE organizes motion experts into tempo-structured groups for different tempo ranges, with multi-scale beat experts capturing fine- and long-range rhythmic dynamics. A Hierarchical Rhythm-Adaptive Routing dynamically selects and fuses experts from music features, enabling flexible, rhythm-aligned generation without manual genre labels. Extensive experiments demonstrate that TempoMoE achieves state-of-the-art results in dance quality and rhythm alignment.
Authors: Ziyuan Tao, Chuanzhi Xu, Sandaru Jayawardana, Wei Bao, Kanchana Thilakarathna, Teng Joon Lim
Abstract: The rapid growth of short-form video platforms increases the need for privacy-preserving moderation, as cloud-based pipelines expose raw videos to privacy risks, high bandwidth costs, and inference latency. To address these challenges, we propose an on-device federated learning framework for video violence detection that integrates self-supervised VideoMAE representations, LoRA-based parameter-efficient adaptation, and defense-in-depth privacy protection. Our approach reduces the trainable parameter count to 5.5M (~3.5% of a 156M backbone) and incorporates DP-SGD with configurable privacy budgets and secure aggregation. Experiments on RWF-2000 with 40 clients achieve 77.25% accuracy without privacy protection and 65-66% under strong differential privacy, while reducing communication cost by $28.3\times$ compared to full-model federated learning. The code is available at: {https://github.com/zyt-599/FedVideoMAE}
Authors: Guangtao Lyu, Xinyi Cheng, Chenghao Xu, Qi Liu, Muli Yang, Fen Fang, Huilin Chen, Jiexi Yan, Xu Yang, Cheng Deng
Abstract: Large Vision-Language Models (LVLMs) have shown remarkable capabilities, yet hallucinations remain a persistent challenge. This work presents a systematic analysis of the internal evolution of visual perception and token generation in LVLMs, revealing two key patterns. First, perception follows a three-stage GATE process: early layers perform a Global scan, intermediate layers Approach and Tighten on core content, and later layers Explore supplementary regions. Second, generation exhibits an SAD (Subdominant Accumulation to Dominant) pattern, where hallucinated tokens arise from the repeated accumulation of subdominant tokens lacking support from attention (visual perception) or feed-forward network (internal knowledge). Guided by these findings, we devise the VDC (Validated Dominance Correction) strategy, which detects unsupported tokens and replaces them with validated dominant ones to improve output reliability. Extensive experiments across multiple models and benchmarks confirm that VDC substantially mitigates hallucinations.
Authors: Yuxiao Yang, Hualian Sheng, Sijia Cai, Jing Lin, Jiahao Wang, Bing Deng, Junzhe Lu, Haoqian Wang, Jieping Ye
Abstract: Video generation models have advanced significantly, yet they still struggle to synthesize complex human movements due to the high degrees of freedom in human articulation. This limitation stems from the intrinsic constraints of pixel-only training objectives, which inherently bias models toward appearance fidelity at the expense of learning underlying kinematic principles. To address this, we introduce EchoMotion, a framework designed to model the joint distribution of appearance and human motion, thereby improving the quality of complex human action video generation. EchoMotion extends the DiT (Diffusion Transformer) framework with a dual-branch architecture that jointly processes tokens concatenated from different modalities. Furthermore, we propose MVS-RoPE (Motion-Video Syncronized RoPE), which offers unified 3D positional encoding for both video and motion tokens. By providing a synchronized coordinate system for the dual-modal latent sequence, MVS-RoPE establishes an inductive bias that fosters temporal alignment between the two modalities. We also propose a Motion-Video Two-Stage Training Strategy. This strategy enables the model to perform both the joint generation of complex human action videos and their corresponding motion sequences, as well as versatile cross-modal conditional generation tasks. To facilitate the training of a model with these capabilities, we construct HuMoVe, a large-scale dataset of approximately 80,000 high-quality, human-centric video-motion pairs. Our findings reveal that explicitly representing human motion is complementary to appearance, significantly boosting the coherence and plausibility of human-centric video generation.
Authors: Hasib Aslam, Muhammad Talal Faiz, Muhammad Imran Malik
Abstract: Advances in neuroscience and artificial intelligence have enabled preliminary decoding of brain activity. However, despite the progress, the interpretability of neural representations remains limited. A significant challenge arises from the intrinsic properties of electroencephalography (EEG) signals, including high noise levels, spatial diffusion, and pronounced temporal variability. To interpret the neural mechanism underlying thoughts, we propose a transformers-based framework to extract spatial-temporal representations associated with observed visual stimuli from EEG recordings. These features are subsequently incorporated into the attention mechanisms of Latent Diffusion Models (LDMs) to facilitate the reconstruction of visual stimuli from brain activity. The quantitative evaluations on publicly available benchmark datasets demonstrate that the proposed method excels at modeling the semantic structures from EEG signals; achieving up to 6.5% increase in latent space clustering accuracy and 11.8% increase in zero shot generalization across unseen classes while having comparable Inception Score and Fr\'echet Inception Distance with existing baselines. Our work marks a significant step towards generalizable semantic interpretation of the EEG signals.
Authors: Sicheng Song, Yanjie Zhang, Zixin Chen, Huamin Qu, Changbo Wang, Chenhui Li
Abstract: The integrity of data visualizations is increasingly threatened by image editing techniques that enable subtle yet deceptive tampering. Through a formative study, we define this challenge and categorize tampering techniques into two primary types: data manipulation and visual encoding manipulation. To address this, we present VizDefender, a framework for tampering detection and analysis. The framework integrates two core components: 1) a semi-fragile watermark module that protects the visualization by embedding a location map to images, which allows for the precise localization of tampered regions while preserving visual quality, and 2) an intent analysis module that leverages Multimodal Large Language Models (MLLMs) to interpret manipulation, inferring the attacker's intent and misleading effects. Extensive evaluations and user studies demonstrate the effectiveness of our methods.
Authors: Alina Elena Baia, Andrea Cavallaro
Abstract: Concept-driven counterfactuals explain decisions of classifiers by altering the model predictions through semantic changes. In this paper, we present a novel approach that leverages cross-modal decompositionality and image-specific concepts to create counterfactual scenarios expressed in natural language. We apply the proposed interpretability framework, termed Decompose and Explain (DeX), to the challenging domain of image privacy decisions, which are contextual and subjective. This application enables the quantification of the differential contributions of key scene elements to the model prediction. We identify relevant decision factors via a multi-criterion selection mechanism that considers both image similarity for minimal perturbations and decision confidence to prioritize impactful changes. This approach evaluates and compares diverse explanations, and assesses the interdependency and mutual influence among explanatory properties. By leveraging image-specific concepts, DeX generates image-grounded, sparse explanations, yielding significant improvements over the state of the art. Importantly, DeX operates as a training-free framework, offering high flexibility. Results show that DeX not only uncovers the principal contributing factors influencing subjective decisions, but also identifies underlying dataset biases allowing for targeted mitigation strategies to improve fairness.
Authors: Maksym Voloshchuk, Bohdana Zarembovska, Mykola Kozlenko
Abstract: Handwritten text recognition and optical character recognition solutions show excellent results with processing data of modern era, but efficiency drops with Latin documents of medieval times. This paper presents a deep learning method to extract text information from handwritten Latin-language documents of the 9th to 11th centuries. The approach takes into account the properties inherent in medieval documents. The paper provides a brief introduction to the field of historical document transcription, a first-sight analysis of the raw data, and the related works and studies. The paper presents the steps of dataset development for further training of the models. The explanatory data analysis of the processed data is provided as well. The paper explains the pipeline of deep learning models to extract text information from the document images, from detecting objects to word recognition using classification models and embedding word images. The paper reports the following results: recall, precision, F1 score, intersection over union, confusion matrix, and mean string distance. The plots of the metrics are also included. The implementation is published on the GitHub repository.
Authors: Kaidi Liang, Ke Li, Xianbiao Hu, Ruwen Qin
Abstract: Automating crash video analysis is essential to leverage the growing availability of driving video data for traffic safety research and accountability attribution in autonomous driving. Crash video analysis is a challenging multitask problem due to the complex spatiotemporal dynamics of crash events in video data and the diverse analytical requirements involved. It requires capabilities spanning crash recognition, temporal grounding, and high-level video understanding. Existing models, however, cannot perform all these tasks within a unified framework, and effective training strategies for such models remain underexplored. To fill these gaps, this paper proposes CrashChat, a multimodal large language model (MLLM) for multitask traffic crash analysis, built upon VideoLLaMA3. CrashChat acquires domain-specific knowledge through instruction fine-tuning and employs a novel multitask learning strategy based on task decoupling and grouping, which maximizes the benefit of joint learning within and across task groups while mitigating negative transfer. Numerical experiments on consolidated public datasets demonstrate that CrashChat consistently outperforms existing MLLMs across model scales and traditional vision-based methods, achieving state-of-the-art performance. It reaches near-perfect accuracy in crash recognition, a 176\% improvement in crash localization, and a 40\% improvement in the more challenging pre-crash localization. Compared to general MLLMs, it substantially enhances textual accuracy and content coverage in crash description and reasoning tasks, with 0.18-0.41 increases in BLEU scores and 0.18-0.42 increases in ROUGE scores. Beyond its strong performance, CrashChat is a convenient, end-to-end analytical tool ready for practical implementation. The dataset and implementation code for CrashChat are available at https://github.com/Liangkd/CrashChat.
Authors: Akshit Achara, Peter Triantafillou, Esther Puyol-Ant\'on, Alexander Hammers, Andrew P. King
Abstract: Deep neural networks often exploit shortcuts. These are spurious cues which are associated with output labels in the training data but are unrelated to task semantics. When the shortcut features are associated with sensitive attributes, shortcut learning can lead to biased model performance. Existing methods for localising and understanding shortcut learning are mostly based upon qualitative, image-level inspection and assume cues are human-visible, limiting their use in domains such as medical imaging. We introduce OSCAR (Ordinal Scoring Correlations for Attribution Representations), a model-agnostic framework for quantifying shortcut learning and localising shortcut features. OSCAR converts image-level task attribution maps into dataset-level rank profiles of image regions and compares them across three models: a balanced baseline model (BA), a test model (TS), and a sensitive attribute predictor (SA). By computing pairwise, partial, and deviation-based correlations on these rank profiles, we produce a set of quantitative metrics that characterise the degree of shortcut reliance for TS, together with a ranking of image-level regions that contribute most to it. Experiments on CelebA, CheXpert, and ADNI show that our correlations are (i) stable across seeds and partitions, (ii) sensitive to the level of association between shortcut features and output labels in the training data, and (iii) able to distinguish localised from diffuse shortcut features. As an illustration of the utility of our method, we show how worst-group performance disparities can be reduced using a simple test-time attenuation approach based on the identified shortcut regions. OSCAR provides a lightweight, pixel-space audit that yields statistical decision rules and spatial maps, enabling users to test, localise, and mitigate shortcut reliance. The code is available at https://github.com/acharaakshit/oscar
Authors: Dmitry Demidov, Zaigham Zaheer, Zongyan Han, Omkar Thawakar, Rao Anwer
Abstract: Vocabulary-free fine-grained image recognition aims to distinguish visually similar categories within a meta-class without a fixed, human-defined label set. Existing solutions for this problem are limited by either the usage of a large and rigid list of vocabularies or by the dependency on complex pipelines with fragile heuristics where errors propagate across stages. Meanwhile, the ability of recent large multi-modal models (LMMs) equipped with explicit or implicit reasoning to comprehend visual-language data, decompose problems, retrieve latent knowledge, and self-correct suggests a more principled and effective alternative. Building on these capabilities, we propose FiNDR (Fine-grained Name Discovery via Reasoning), the first reasoning-augmented LMM-based framework for vocabulary-free fine-grained recognition. The system operates in three automated steps: (i) a reasoning-enabled LMM generates descriptive candidate labels for each image; (ii) a vision-language model filters and ranks these candidates to form a coherent class set; and (iii) the verified names instantiate a lightweight multi-modal classifier used at inference time. Extensive experiments on popular fine-grained classification benchmarks demonstrate state-of-the-art performance under the vocabulary-free setting, with a significant relative margin of up to 18.8% over previous approaches. Remarkably, the proposed method surpasses zero-shot baselines that exploit pre-defined ground-truth names, challenging the assumption that human-curated vocabularies define an upper bound. Additionally, we show that carefully curated prompts enable open-source LMMs to match proprietary counterparts. These findings establish reasoning-augmented LMMs as an effective foundation for scalable, fully automated, open-world fine-grained visual recognition. The source code is available on github.com/demidovd98/FiNDR.
Authors: Mohamad Zamini, Diksha Shukla
Abstract: Multimodal Large Language Models (MLLMs) combine visual and textual representations to enable rich reasoning capabilities. However, the high computational cost of processing dense visual tokens remains a major bottleneck. A critical component in this pipeline is the visual projector, which bridges the vision encoder and the language model. Standard designs often employ a simple multi-layer perceptron for direct token mapping, but this approach scales poorly with high-resolution inputs, introducing significant redundancy. We present Delta-LLaVA, a token-efficient projector that employs a low-rank DeltaProjection to align multi-level vision features into a compact subspace before further interaction. On top of this base alignment, lightweight Transformer blocks act as specialization layers, capturing both global and local structure under constrained token budgets. Extensive experiments and ablations demonstrate that this base-then-specialize design yields consistent gains across multiple benchmarks with only 144 tokens, highlighting the importance of token formation prior to scaling interaction capacity. With Delta-LLaVA, inference throughput improves by up to 55%, while end-to-end training accelerates by nearly 4-5x in pretraining and over 1.5x in finetuning, highlighting the dual benefits of our design in both efficiency and scalability.
Authors: Raina Panda, Daniel Fein, Arpita Singhal, Mark Fiore, Maneesh Agrawala, Matyas Bohacek
Abstract: Artistic style transfer in generative models remains a significant challenge, as existing methods often introduce style only via model fine-tuning, additional adapters, or prompt engineering, all of which can be computationally expensive and may still entangle style with subject matter. In this paper, we introduce a training- and inference-light, interpretable method for representing and transferring artistic style. Our approach leverages an art-specific Sparse Autoencoder (SAE) on top of latent embeddings of generative image models. Trained on artistic data, our SAE learns an emergent, largely disentangled set of stylistic and compositional concepts, corresponding to style-related elements pertaining brushwork, texture, and color palette, as well as semantic and structural concepts. We call it LouvreSAE and use it to construct style profiles: compact, decomposable steering vectors that enable style transfer without any model updates or optimization. Unlike prior concept-based style transfer methods, our method requires no fine-tuning, no LoRA training, and no additional inference passes, enabling direct steering of artistic styles from only a few reference images. We validate our method on ArtBench10, achieving or surpassing existing methods on style evaluations (VGG Style Loss and CLIP Score Style) while being 1.7-20x faster and, critically, interpretable.
Authors: Hang Yu, Juntu Zhao, Yufeng Liu, Kaiyu Li, Cheng Ma, Di Zhang, Yingdong Hu, Guang Chen, Junyuan Xie, Junliang Guo, Junqiao Zhao, Yang Gao
Abstract: Vision-Language-Action (VLA) models align vision and language with embodied control, but their object referring ability remains limited when relying solely on text prompt, especially in cluttered or out-of-distribution (OOD) scenes. In this study, we introduce the Point-VLA, a plug-and-play policy that augments language instructions with explicit visual cues (e.g., bounding boxes) to resolve referential ambiguity and enable precise object-level grounding. To efficiently scale visually grounded datasets, we further develop an automatic data annotation pipeline requiring minimal human effort. We evaluate Point-VLA on diverse real-world referring tasks and observe consistently stronger performance than text-only instruction VLAs, particularly in cluttered or unseen-object scenarios, with robust generalization. These results demonstrate that Point-VLA effectively resolves object referring ambiguity through pixel-level visual grounding, achieving more generalizable embodied control.
Authors: Nicolas Caytuiro, Ivan Sipiran
Abstract: We propose a novel data-centric approach to promote symmetry in 3D generative models by modifying the training data rather than the model architecture. Our method begins with an analysis of reflectional symmetry in both real-world 3D shapes and samples generated by state-of-the-art models. We hypothesize that training a generative model exclusively on half-objects, obtained by reflecting one half of the shapes along the x=0 plane, enables the model to learn a rich distribution of partial geometries which, when reflected during generation, yield complete shapes that are both visually plausible and geometrically symmetric. To test this, we construct a new dataset of half-objects from three ShapeNet classes (Airplane, Car, and Chair) and train two generative models. Experiments demonstrate that the generated shapes are symmetrical and consistent, compared with the generated objects from the original model and the original dataset objects.
Authors: Zaidao Han, Risa Higashita, Jiang Liu
Abstract: Camera-based 3D Semantic Scene Completion (SSC) is a critical task for autonomous driving and robotic scene understanding. It aims to infer a complete 3D volumetric representation of both semantics and geometry from a single image. Existing methods typically focus on end-to-end 2D-to-3D feature lifting and voxel completion. However, they often overlook the interference between high-confidence visible-region perception and low-confidence occluded-region reasoning caused by single-image input, which can lead to feature dilution and error propagation. To address these challenges, we introduce an offline Visible Region Label Extraction (VRLE) strategy that explicitly separates and extracts voxel-level supervision for visible regions from dense 3D ground truth. This strategy purifies the supervisory space for two complementary sub-tasks: visible-region perception and occluded-region reasoning. Building on this idea, we propose the Visible-Occluded Interactive Completion Network (VOIC), a novel dual-decoder framework that explicitly decouples SSC into visible-region semantic perception and occluded-region scene completion. VOIC first constructs a base 3D voxel representation by fusing image features with depth-derived occupancy. The visible decoder focuses on generating high-fidelity geometric and semantic priors, while the occlusion decoder leverages these priors together with cross-modal interaction to perform coherent global scene reasoning. Extensive experiments on the SemanticKITTI and SSCBench-KITTI360 benchmarks demonstrate that VOIC outperforms existing monocular SSC methods in both geometric completion and semantic segmentation accuracy, achieving state-of-the-art performance.
Authors: Guandong Li, Yijun Ding
Abstract: Recent tuning-free identity customization methods achieve high facial fidelity but often overlook visual context, such as lighting, skin texture, and environmental tone. This limitation leads to ``Semantic-Visual Dissonance,'' where accurate facial geometry clashes with the input's unique atmosphere, causing an unnatural ``sticker-like'' effect. We propose **DVI (Disentangled Visual-Identity)**, a zero-shot framework that orthogonally disentangles identity into fine-grained semantic and coarse-grained visual streams. Unlike methods relying solely on semantic vectors, DVI exploits the inherent statistical properties of the VAE latent space, utilizing mean and variance as lightweight descriptors for global visual atmosphere. We introduce a **Parameter-Free Feature Modulation** mechanism that adaptively modulates semantic embeddings with these visual statistics, effectively injecting the reference's ``visual soul'' without training. Furthermore, a **Dynamic Temporal Granularity Scheduler** aligns with the diffusion process, prioritizing visual atmosphere in early denoising stages while refining semantic details later. Extensive experiments demonstrate that DVI significantly enhances visual consistency and atmospheric fidelity without parameter fine-tuning, maintaining robust identity preservation and outperforming state-of-the-art methods in IBench evaluations.
Authors: Tianle Lu, Ke Chen, Yuping Duan
Abstract: We introduce a novel formulation for curvature regularization by penalizing normal curvatures from multiple directions. This total normal curvature regularization is capable of producing solutions with sharp edges and precise isotropic properties. To tackle the resulting high-order nonlinear optimization problem, we reformulate it as the task of finding the steady-state solution of a time-dependent partial differential equation (PDE) system. Time discretization is achieved through operator splitting, where each subproblem at the fractional steps either has a closed-form solution or can be efficiently solved using advanced algorithms. Our method circumvents the need for complex parameter tuning and demonstrates robustness to parameter choices. The efficiency and effectiveness of our approach have been rigorously validated in the context of surface and image smoothing problems.
Authors: Cheng-Hong Chang, Pei-Hsuan Tsai
Abstract: Object recognition has become prevalent across various industries. However, most existing applications are limited to identifying objects alone, without considering their associated states. The ability to recognize both the state and object simultaneously remains less common. One approach to address this is by treating state and object as a single category during training. However, this approach poses challenges in data collection and training since it requires comprehensive data for all possible combinations. Compositional Zero-shot Learning (CZSL) emerges as a viable solution by treating the state and object as distinct categories during training. CZSL facilitates the identification of novel compositions even in the absence of data for every conceivable combination. The current state-of-the-art method, KG-SP, addresses this issue by training distinct classifiers for states and objects, while leveraging a semantic model to evaluate the plausibility of composed compositions. However, KG-SP's accuracy in state and object recognition can be further improved, and it fails to consider the weighting of states and objects during composition. In this study, we propose SASOW, an enhancement of KG-SP that considers the weighting of states and objects while improving composition recognition accuracy. First, we introduce self-attention mechanisms into the classifiers for states and objects, leading to enhanced accuracy in recognizing both. Additionally, we incorporate the weighting of states and objects during composition to generate more reasonable and accurate compositions. Our validation process involves testing SASOW on three established benchmark datasets. Experimental outcomes affirm when compared against OW-CZSL approach, KG-SP, SASOW showcases improvements of 2.1%, 1.7%, and 0.4% in terms of accuracy for unseen compositions across the MIT-States, UT Zappos, and C-GQA datasets, respectively.
Authors: Gyeongrok Oh, Youngdong Jang, Jonghyun Choi, Suk-Ju Kang, Guang Lin, Sangpil Kim
Abstract: Dominant paradigms for 4D LiDAR panoptic segmentation are usually required to train deep neural networks with large superimposed point clouds or design dedicated modules for instance association. However, these approaches perform redundant point processing and consequently become computationally expensive, yet still overlook the rich geometric priors inherently provided by raw point clouds. To this end, we introduce ICP-4D, a simple yet effective training-free framework that unifies spatial and temporal reasoning through geometric relations among instance-level point sets. Specifically, we apply the Iterative Closest Point (ICP) algorithm to directly associate temporally consistent instances by aligning the source and target point sets through the estimated transformation. To stabilize association under noisy instance predictions, we introduce a Sinkhorn-based soft matching. This exploits the underlying instance distribution to obtain accurate point-wise correspondences, resulting in robust geometric alignment. Furthermore, our carefully designed pipeline, which considers three instance types-static, dynamic, and missing-offers computational efficiency and occlusion-aware matching. Our extensive experiments across both SemanticKITTI and panoptic nuScenes demonstrate that our method consistently outperforms state-of-the-art approaches, even without additional training or extra point cloud inputs.
Authors: Cheng Yaw Low, Heejoon Koo, Jaewoo Park, Kaleb Mesfin Asfaw, Meeyoung Cha
Abstract: AI-guided classification of ecological families, genera, and species underpins global sustainability efforts such as biodiversity monitoring, conservation planning, and policy-making. Progress toward this goal is hindered by long-tailed taxonomic distributions from class imbalance, along with fine-grained taxonomic variations, test-time spatiotemporal domain shifts, and closed-set assumptions that can only recognize previously seen taxa. We introduce the Open-World Ecological Taxonomy Classification, a unified framework that captures the co-occurrence of these challenges in realistic ecological settings. To address them, we propose TaxoNet, an embedding-based encoder with a dual-margin penalization loss that strengthens learning signals from rare underrepresented taxa while mitigating the dominance of overrepresented ones, directly confronting interrelated challenges. We evaluate our method on diverse ecological domains: Google Auto-Arborist (urban trees), iNat-Plantae (Plantae observations from various ecosystems in iNaturalist-2019), and NAFlora-Mini (a curated herbarium collection). Our model consistently outperforms baselines, particularly for rare taxa, establishing a strong foundation for open-world plant taxonomic monitoring. Our findings further show that general-purpose multimodal foundation models remain constrained in plant-domain applications.
Authors: Zelin Zhao, Xinyu Gong, Bangya Liu, Ziyang Song, Jun Zhang, Suhui Wu, Yongxin Chen, Hao Zhang
Abstract: Achieving precise camera control in video generation remains challenging, as existing methods often rely on camera pose annotations that are difficult to scale to large and dynamic datasets and are frequently inconsistent with depth estimation, leading to train-test discrepancies. We introduce CETCAM, a camera-controllable video generation framework that eliminates the need for camera annotations through a consistent and extensible tokenization scheme. CETCAM leverages recent advances in geometry foundation models, such as VGGT, to estimate depth and camera parameters and converts them into unified, geometry-aware tokens. These tokens are seamlessly integrated into a pretrained video diffusion backbone via lightweight context blocks. Trained in two progressive stages, CETCAM first learns robust camera controllability from diverse raw video data and then refines fine-grained visual quality using curated high-fidelity datasets. Extensive experiments across multiple benchmarks demonstrate state-of-the-art geometric consistency, temporal stability, and visual realism. Moreover, CETCAM exhibits strong adaptability to additional control modalities, including inpainting and layout control, highlighting its flexibility beyond camera control. The project page is available at https://sjtuytc.github.io/CETCam_project_page.github.io/.
URLs: https://sjtuytc.github.io/CETCam_project_page.github.io/.
Authors: Sihao Lin, Zerui Li, Xunyi Zhao, Gengze Zhou, Liuyi Wang, Rong Wei, Rui Tang, Juncheng Li, Hanqing Wang, Jiangmiao Pang, Anton van den Hengel, Jiajun Liu, Qi Wu
Abstract: Despite remarkable progress in Vision-Language Navigation (VLN), existing benchmarks remain confined to fixed, small-scale datasets with naive physical simulation. These shortcomings limit the insight that the benchmarks provide into sim-to-real generalization, and create a significant research gap. Furthermore, task fragmentation prevents unified/shared progress in the area, while limited data scales fail to meet the demands of modern LLM-based pretraining. To overcome these limitations, we introduce VLNVerse: a new large-scale, extensible benchmark designed for Versatile, Embodied, Realistic Simulation, and Evaluation. VLNVerse redefines VLN as a scalable, full-stack embodied AI problem. Its Versatile nature unifies previously fragmented tasks into a single framework and provides an extensible toolkit for researchers. Its Embodied design moves beyond intangible and teleporting "ghost" agents that support full-kinematics in a Realistic Simulation powered by a robust physics engine. We leverage the scale and diversity of VLNVerse to conduct a comprehensive Evaluation of existing methods, from classic models to MLLM-based agents. We also propose a novel unified multi-task model capable of addressing all tasks within the benchmark. VLNVerse aims to narrow the gap between simulated navigation and real-world generalization, providing the community with a vital tool to boost research towards scalable, general-purpose embodied locomotion agents.
Authors: Haoze Li, Jie Zhang, Guoying Zhao, Stephen Lin, Shiguang Shan
Abstract: Face Presentation Attack Detection (PAD) demands incremental learning (IL) to combat evolving spoofing tactics and domains. Privacy regulations, however, forbid retaining past data, necessitating rehearsal-free IL (RF-IL). Vision-Language Pre-trained (VLP) models, with their prompt-tunable cross-modal representations, enable efficient adaptation to new spoofing styles and domains. Capitalizing on this strength, we propose \textbf{SVLP-IL}, a VLP-based RF-IL framework that balances stability and plasticity via \textit{Multi-Aspect Prompting} (MAP) and \textit{Selective Elastic Weight Consolidation} (SEWC). MAP isolates domain dependencies, enhances distribution-shift sensitivity, and mitigates forgetting by jointly exploiting universal and domain-specific cues. SEWC selectively preserves critical weights from previous tasks, retaining essential knowledge while allowing flexibility for new adaptations. Comprehensive experiments across multiple PAD benchmarks show that SVLP-IL significantly reduces catastrophic forgetting and enhances performance on unseen domains. SVLP-IL offers a privacy-compliant, practical solution for robust lifelong PAD deployment in RF-IL settings.
Authors: Connor Kilrain, David Carlyn, Julia Chae, Sara Beery, Wei-Lun Chao, Jianyang Gu
Abstract: The rise of personalized generative models raises a central question: how should we evaluate identity preservation? Given a reference image (e.g., one's pet), we expect the generated image to retain precise details attached to the subject's identity. However, current generative evaluation metrics emphasize the overall semantic similarity between the reference and the output, and overlook these fine-grained discriminative details. We introduce Finer-Personalization Rank, an evaluation protocol tailored to identity preservation. Instead of pairwise similarity, Finer-Personalization Rank adopts a ranking view: it treats each generated image as a query against an identity-labeled gallery consisting of visually similar real images. Retrieval metrics (e.g., mean average precision) measure performance, where higher scores indicate that identity-specific details (e.g., a distinctive head spot) are preserved. We assess identity at multiple granularities -- from fine-grained categories (e.g., bird species, car models) to individual instances (e.g., re-identification). Across CUB, Stanford Cars, and animal Re-ID benchmarks, Finer-Personalization Rank more faithfully reflects identity retention than semantic-only metrics and reveals substantial identity drift in several popular personalization methods. These results position the gallery-based protocol as a principled and practical evaluation for personalized generation.
Authors: Ran Li, Pan Xiao, Kaushik Dutta, Youdong Guo
Abstract: Fluorescence Microcopy Calcium Imaging is a fundamental tool to in-vivo record and analyze large scale neuronal activities simultaneously at a single cell resolution. Automatic and precise detection of behaviorally relevant neuron activity from the recordings is critical to study the mapping of brain activity in organisms. However a perpetual bottleneck to this problem is the manual segmentation which is time and labor intensive and lacks generalizability. To this end, we present a Bayesian Deep Learning Framework to detect neuronal activities in 4D spatio-temporal data obtained by light sheet microscopy. Our approach accounts for the use of temporal information by calculating pixel wise correlation maps and combines it with spatial information given by the mean summary image. The Bayesian framework not only produces probability segmentation maps but also models the uncertainty pertaining to active neuron detection. To evaluate the accuracy of our framework we implemented the test of reproducibility to assert the generalization of the network to detect neuron activity. The network achieved a mean Dice Score of 0.81 relative to the synthetic Ground Truth obtained by Otsu's method and a mean Dice Score of 0.79 between the first and second run for test of reproducibility. Our method successfully deployed can be used for rapid detection of active neuronal activities for behavioural studies.
Authors: Xiaoyang Li, Mingming Lu, Ruiqi Wang, Hao Li, Zewei Le
Abstract: Few-shot action recognition aims to enable models to quickly learn new action categories from limited labeled samples, addressing the challenge of data scarcity in real-world applications. Current research primarily addresses three core challenges: (1) temporal modeling, where models are prone to interference from irrelevant static background information and struggle to capture the essence of dynamic action features; (2) visual similarity, where categories with subtle visual differences are difficult to distinguish; and (3) the modality gap between visual-textual support prototypes and visual-only queries, which complicates alignment within a shared embedding space. To address these challenges, this paper proposes a CLIP-SPM framework, which includes three components: (1) the Hierarchical Synergistic Motion Refinement (HSMR) module, which aligns deep and shallow motion features to improve temporal modeling by reducing static background interference; (2) the Semantic Prototype Modulation (SPM) strategy, which generates query-relevant text prompts to bridge the modality gap and integrates them with visual features, enhancing the discriminability between similar actions; and (3) the Prototype-Anchor Dual Modulation (PADM) method, which refines support prototypes and aligns query features with a global semantic anchor, improving consistency across support and query samples. Comprehensive experiments across standard benchmarks, including Kinetics, SSv2-Full, SSv2-Small, UCF101, and HMDB51, demonstrate that our CLIP-SPM achieves competitive performance under 1-shot, 3-shot, and 5-shot settings. Extensive ablation studies and visual analyses further validate the effectiveness of each component and its contributions to addressing the core challenges. The source code and models are publicly available at GitHub.
Authors: Utae Jeong, Sumin In, Hyunju Ryu, Jaewan Choi, Feng Yang, Jongheon Jeong, Seungryong Kim, Sangpil Kim
Abstract: Image watermarking supports authenticity and provenance, yet many schemes are still easy to bypass with various distortions and powerful generative edits. Deep learning-based watermarking has improved robustness to diffusion-based image editing, but a gap remains when a watermarked image is converted to video by image-to-video (I2V), in which per-frame watermark detection weakens. I2V has quickly advanced from short, jittery clips to multi-second, temporally coherent scenes, and it now serves not only content creation but also world-modeling and simulation workflows, making cross-modal watermark recovery crucial. We present WaTeRFlow, a framework tailored for robustness under I2V. It consists of (i) FUSE (Flow-guided Unified Synthesis Engine), which exposes the encoder-decoder to realistic distortions via instruction-driven edits and a fast video diffusion proxy during training, (ii) optical-flow warping with a Temporal Consistency Loss (TCL) that stabilizes per-frame predictions, and (iii) a semantic preservation loss that maintains the conditioning signal. Experiments across representative I2V models show accurate watermark recovery from frames, with higher first-frame and per-frame bit accuracy and resilience when various distortions are applied before or after video generation.
Authors: Hwanhee Jung, Seunggwan Lee, Jeongyoon Yoon, SeungHyeon Kim, Giljoo Nam, Qixing Huang, Sangpil Kim
Abstract: Synthesizing realistic human-object interaction (HOI) is essential for 3D computer vision and robotics, underpinning animation and embodied control. Existing approaches often require manually specified intermediate waypoints and place all optimization objectives on a single network, which increases complexity, reduces flexibility, and leads to errors such as unsynchronized human and object motion or penetration. To address these issues, we propose Decoupled Generative Modeling for Human-Object Interaction Synthesis (DecHOI), which separates path planning and action synthesis. A trajectory generator first produces human and object trajectories without prescribed waypoints, and an action generator conditions on these paths to synthesize detailed motions. To further improve contact realism, we employ adversarial training with a discriminator that focuses on the dynamics of distal joints. The framework also models a moving counterpart and supports responsive, long-sequence planning in dynamic scenes, while preserving plan consistency. Across two benchmarks, FullBodyManipulation and 3D-FUTURE, DecHOI surpasses prior methods on most quantitative metrics and qualitative evaluations, and perceptual studies likewise prefer our results.
Authors: Jihui Guo, Zongmin Zhang, Zhen Sun, Yuhao Yang, Jinlin Wu, Fu Zhang, Xinlei He
Abstract: Deep learning advances have enabled accurate six-degree-of-freedom (6DoF) object pose estimation, widely used in robotics, AR/VR, and autonomous systems. However, backdoor attacks pose significant security risks. While most research focuses on 2D vision, 6DoF pose estimation remains largely unexplored. Unlike traditional backdoors that only change classes, 6DoF attacks must control continuous parameters like translation and rotation, rendering 2D methods inapplicable. We propose 6DAttack, a framework using 3D object triggers to induce controlled erroneous poses while maintaining normal behavior. Evaluations on PVNet, DenseFusion, and PoseDiffusion across LINEMOD, YCB-Video, and CO3D show high attack success rates (ASRs) without compromising clean performance. Backdoored models achieve up to 100% clean ADD accuracy and 100% ASR, with triggered samples reaching 97.70% ADD-P. Furthermore, a representative defense remains ineffective. Our findings reveal a serious, underexplored threat to 6DoF pose estimation.
Authors: Ruiqi Ma, Yu Yan, Chunhong Zhang, Minghao Yin, XinChao Liu, Zhihong Jin, Zheng Hu
Abstract: Large Vision-Language Models (LVLMs) bridge the gap between visual and linguistic modalities, demonstrating strong potential across a variety of domains. However, despite significant progress, LVLMs still suffer from severe hallucination issues in object recognition tasks. These models often fail to accurately identify certain objects, leading to text generation that appears fluent but does not correspond to the visual content, which can have serious consequences in real-world applications. Recently, several methods have been proposed to alleviate LVLM hallucinations, but most focus solely on reducing hallucinations in the language modality. To mitigate hallucinations in both the language and visual modalities, we introduce Hallucination Disentangled Decoding (HDD) method that requires no training. HDD enhances the original image by segmenting it and selecting images that augment the original, while also utilizing a blank image to eliminate language prior hallucinations in both the original and segmented images. This design not only reduces the model's dependence on language priors but also enhances its visual performance. (Code: https://github.com/rickeyhhh/Hallucination-Disentangled-Decoding)
URLs: https://github.com/rickeyhhh/Hallucination-Disentangled-Decoding)
Authors: Khanh Nguyen, Dasith de Silva Edirimuni, Ghulam Mubashar Hassan, Ajmal Mian
Abstract: Locating and retrieving objects from scene-level point clouds is a challenging problem with broad applications in robotics and augmented reality. This task is commonly formulated as open-vocabulary 3D instance segmentation. Although recent methods demonstrate strong performance, they depend heavily on SAM and CLIP to generate and classify 3D instance masks from images accompanying the point cloud, leading to substantial computational overhead and slow processing that limit their deployment in real-world settings. Open-YOLO 3D alleviates this issue by using a real-time 2D detector to classify class-agnostic masks produced directly from the point cloud by a pretrained 3D segmenter, eliminating the need for SAM and CLIP and significantly reducing inference time. However, Open-YOLO 3D often fails to generalize to object categories that appear infrequently in the 3D training data. In this paper, we propose a method that generates 3D instance masks for novel objects from RGB images guided by a 2D open-vocabulary detector. Our approach inherits the 2D detector's ability to recognize novel objects while maintaining efficient classification, enabling fast and accurate retrieval of rare instances from open-ended text queries. Our code will be made available at https://github.com/ndkhanh360/BoxOVIS.
Authors: Ariel Lubonja, Pedro R. A. S. Bassi, Wenxuan Li, Hualin Qiao, Randal Burns, Alan L. Yuille, Zongwei Zhou
Abstract: Open challenges have become the de facto standard for comparative ranking of medical AI methods. Despite their importance, medical AI leaderboards exhibit three persistent limitations: (1) score gaps are rarely tested for statistical significance, so rank stability is unknown; (2) single averaged metrics are applied to every organ, hiding clinically important boundary errors; (3) performance across intersecting demographics is seldom reported, masking fairness and equity gaps. We introduce RankInsight, an open-source toolkit that seeks to address these limitations. RankInsight (1) computes pair-wise significance maps that show the nnU-Net family outperforms Vision-Language and MONAI submissions with high statistical certainty; (2) recomputes leaderboards with organ-appropriate metrics, reversing the order of the top four models when Dice is replaced by NSD for tubular structures; and (3) audits intersectional fairness, revealing that more than half of the MONAI-based entries have the largest gender-race discrepancy on our proprietary Johns Hopkins Hospital dataset. The RankInsight toolkit is publicly released and can be directly applied to past, ongoing, and future challenges. It enables organizers and participants to publish rankings that are statistically sound, clinically meaningful, and demographically fair.
Authors: Weiyi Lyu, Xinming Fang, Jun Wang, Jun Shi, Guixu Zhang, Juncheng Li
Abstract: Magnetic resonance imaging (MRI) is a cornerstone of modern clinical diagnosis, offering unparalleled soft-tissue contrast without ionizing radiation. However, prolonged scan times remain a major barrier to patient throughput and comfort. Existing accelerated MRI techniques often struggle with two key challenges: (1) failure to effectively utilize inherent K-space prior information, leading to persistent aliasing artifacts from zero-filled inputs; and (2) contamination of target reconstruction quality by irrelevant information when employing multi-contrast fusion strategies. To overcome these challenges, we present MambaMDN, a dual-domain framework for multi-contrast MRI reconstruction. Our approach first employs fully-sampled reference K-space data to complete the undersampled target data, generating structurally aligned but modality-mixed inputs. Subsequently, we develop a Mamba-based modality disentanglement network to extract and remove reference-specific features from the mixed representation. Furthermore, we introduce an iterative refinement mechanism to progressively enhance reconstruction accuracy through repeated feature purification. Extensive experiments demonstrate that MambaMDN can significantly outperform existing multi-contrast reconstruction methods.
Authors: Tiantian Li, Xinjie Zhang, Xingtong Ge, Tongda Xu, Dailan He, Jun Zhang, Yan Wang
Abstract: Implicit neural representations (INRs) have achieved remarkable success in image representation and compression, but they require substantial training time and memory. Meanwhile, recent 2D Gaussian Splatting (GS) methods (\textit{e.g.}, GaussianImage) offer promising alternatives through efficient primitive-based rendering. However, these methods require excessive Gaussian primitives to maintain high visual fidelity. To exploit the potential of GS-based approaches, we present GaussianImage++, which utilizes limited Gaussian primitives to achieve impressive representation and compression performance. Firstly, we introduce a distortion-driven densification mechanism. It progressively allocates Gaussian primitives according to signal intensity. Secondly, we employ context-aware Gaussian filters for each primitive, which assist in the densification to optimize Gaussian primitives based on varying image content. Thirdly, we integrate attribute-separated learnable scalar quantizers and quantization-aware training, enabling efficient compression of primitive attributes. Experimental results demonstrate the effectiveness of our method. In particular, GaussianImage++ outperforms GaussianImage and INRs-based COIN in representation and compression performance while maintaining real-time decoding and low memory usage.
Authors: Tao Li, Zhenbao Yu, Banglei Guan, Jianli Han, Weimin Lv, Friedrich Fraundorfer
Abstract: This work presents two novel solvers for estimating the relative poses among views with known vertical directions. The vertical directions of camera views can be easily obtained using inertial measurement units (IMUs) which have been widely used in autonomous vehicles, mobile phones, and unmanned aerial vehicles (UAVs). Given the known vertical directions, our lgorithms only need to solve for two rotation angles and two translation vectors. In this paper, a linear closed-form solution has been described, requiring only four point correspondences in three views. We also propose a minimal solution with three point correspondences using the latest Gr\"obner basis solver. Since the proposed methods require fewer point correspondences, they can be efficiently applied within the RANSAC framework for outliers removal and pose estimation in visual odometry. The proposed method has been tested on both synthetic data and real-world scenes from KITTI. The experimental results show that the accuracy of the estimated poses is superior to other alternative methods.
Authors: Hengyi Feng, Zeang Sheng, Meiyi Qiang, Wentao Zhang
Abstract: Despite the remarkable success of multimodal large language models (MLLMs) in generative tasks, we observe that they exhibit a counterintuitive deficiency in the zero-shot multimodal retrieval task. In this work, we investigate the underlying mechanisms that hinder MLLMs from serving as effective retrievers. With the help of sparse autoencoders (SAEs), we decompose MLLM output representations into interpretable semantic concepts to probe their intrinsic behavior. Our analysis reveals that the representation space of MLLMs is overwhelmingly dominated by textual semantics; the visual information essential for multimodal retrieval only constitutes a small portion. This imbalance is compounded by the heavy focus of MLLMs on bridging image-text modalities, which facilitates generation but homogenizes embeddings and finally diminishes the discriminative power required for multimodal retrieval. We further discover that the specific feature components that contribute most to the similarity computations for MLLMs are in fact distractors that actively degrade retrieval performance. Overall, our work provides the first in-depth interpretability analysis of MLLM representations in the context of multimodal retrieval and offers possible directions for enhancing the multimodal retrieval capabilities of MLLMs.
Authors: Ruikai Li, Xinrun Li, Mengwei Xie, Hao Shan, Shoumeng Qiu, Xinyuan Chang, Yizhe Fan, Feng Xiong, Han Jiang, Yilong Ren, Haiyang Yu, Mu Xu, Yang Long, Varun Ojha, Zhiyong Cui
Abstract: Online High-Definition (HD) map construction is pivotal for autonomous driving. While recent approaches leverage historical temporal fusion to improve performance, we identify a critical safety flaw in this paradigm: it is inherently ``spatially backward-looking." These methods predominantly enhance map reconstruction in traversed areas, offering minimal improvement for the unseen road ahead. Crucially, our analysis of downstream planning tasks reveals a severe asymmetry: while rearward perception errors are often tolerable, inaccuracies in the forward region directly precipitate hazardous driving maneuvers. To bridge this safety gap, we propose AMap, a novel framework for Ahead-aware online HD Mapping. We pioneer a ``distill-from-future" paradigm, where a teacher model with privileged access to future temporal contexts guides a lightweight student model restricted to the current frame. This process implicitly compresses prospective knowledge into the student model, endowing it with ``look-ahead" capabilities at zero inference-time cost. Technically, we introduce a Multi-Level BEV Distillation strategy with spatial masking and an Asymmetric Query Adaptation module to effectively transfer future-aware representations to the student's static queries. Extensive experiments on the nuScenes and Argoverse 2 benchmark demonstrate that AMap significantly enhances current-frame perception. Most notably, it outperforms state-of-the-art temporal models in critical forward regions while maintaining the efficiency of single current frame inference.
Authors: Wendong Bu, Kaihang Pan, Yuze Lin, Jiacheng Li, Kai Shen, Wenqiao Zhang, Juncheng Li, Jun Xiao, Siliang Tang
Abstract: Large language models (LLMs) have unified diverse linguistic tasks within a single framework, yet such unification remains unexplored in human motion generation. Existing methods are confined to isolated tasks, limiting flexibility for free-form and omni-objective generation. To address this, we propose OmniMoGen, a unified framework that enables versatile motion generation through interleaved text-motion instructions. Built upon a concise RVQ-VAE and transformer architecture, OmniMoGen supports end-to-end instruction-driven motion generation. We construct X2Mo, a large-scale dataset of over 137K interleaved text-motion instructions, and introduce AnyContext, a benchmark for evaluating interleaved motion generation. Experiments show that OmniMoGen achieves state-of-the-art performance on text-to-motion, motion editing, and AnyContext, exhibiting emerging capabilities such as compositional editing, self-reflective generation, and knowledge-informed generation. These results mark a step toward the next intelligent motion generation. Project Page: https://OmniMoGen.github.io/.
Authors: Marios Thoma (CYENS Centre of Excellence, Nicosia, Cyprus, Open University Cyprus, Nicosia, Cyprus), Zenonas Theodosiou (CYENS Centre of Excellence, Nicosia, Cyprus, Department of Communication and Internet Studies, Cyprus University of Technology, Limassol, Cyprus), Harris Partaourides (AI Cyprus Ethical Novelties Ltd, Limassol, Cyprus), Vassilis Vassiliades (CYENS Centre of Excellence, Nicosia, Cyprus), Loizos Michael (Open University Cyprus, Nicosia, Cyprus, CYENS Centre of Excellence, Nicosia, Cyprus), Andreas Lanitis (CYENS Centre of Excellence, Nicosia, Cyprus, Department of Multimedia and Graphic Arts, Cyprus University of Technology, Limassol, Cyprus)
Abstract: Walking has always been a primary mode of transportation and is recognized as an essential activity for maintaining good health. Despite the need for safe walking conditions in urban environments, sidewalks are frequently obstructed by various obstacles that hinder free pedestrian movement. Any object obstructing a pedestrian's path can pose a safety hazard. The advancement of pervasive computing and egocentric vision techniques offers the potential to design systems that can automatically detect such obstacles in real time, thereby enhancing pedestrian safety. The development of effective and efficient identification algorithms relies on the availability of comprehensive and well-balanced datasets of egocentric data. In this work, we introduce the PEDESTRIAN dataset, comprising egocentric data for 29 different obstacles commonly found on urban sidewalks. A total of 340 videos were collected using mobile phone cameras, capturing a pedestrian's point of view. Additionally, we present the results of a series of experiments that involved training several state-of-the-art deep learning algorithms using the proposed dataset, which can be used as a benchmark for obstacle detection and recognition tasks. The dataset can be used for training pavement obstacle detectors to enhance the safety of pedestrians in urban areas.
Authors: Zihao Luo, Shaohao Rui, Zhenyu Tang, Guotai Wang, Xiaosong Wang
Abstract: Continual self-supervised learning (CSSL) in medical imaging trains a foundation model sequentially, alleviating the need for collecting multi-modal images for joint training and offering promising improvements in downstream performance while preserving data privacy. However, most existing methods still rely on replaying data from previous stages to prevent catastrophic forgetting, which compromises privacy and limits their applicability in real-world scenarios where data transfer across sites is often restricted. In this work, we propose InvCoSS, an inversion-driven continual self-supervised learning framework for medical multi-modal image pre-training. Specifically, after training on a previous task, InvCoSS inverts the pre-trained self-supervised model to generate synthetic images that approximate the original training distribution. These synthetic images are then combined with data from the new task for joint optimization, which effectively mitigates catastrophic forgetting while strictly adhering to the constraint of no access to previous real data. Furthermore, to improve the fidelity of synthetic images, we introduce a novel InvUNet with a multi-scale fusion architecture to restore both high- and low-frequency components of the inverted images. To enhance diversity and prevent mode collapse, we design a repulsive representation-learning mechanism that encourages a diverse feature space for synthetic images without class guidance. Extensive experiments across nine downstream tasks validate the effectiveness of InvCoSS, achieving performance comparable to or even superior to prior data-replay methods while significantly reducing storage requirements and eliminating data privacy constraints.
Authors: Na Gao, Chenfei Ye, Yanwu Yang, Anqi Li, Zhengbo He, Li Liang, Zhiyuan Liu, Xingyu Hao, Ting Ma, Tengfei Guo
Abstract: Accurate characterization of hippocampal substructure is crucial for detecting subtle structural changes and identifying early neurodegenerative biomarkers. However, high inter-subject variability and complex folding pattern of human hippocampus hinder consistent cross-subject and longitudinal analysis. Most existing approaches rely on subject-specific modelling and lack a stable intrinsic coordinate system to accommodate anatomical variability, which limits their ability to establish reliable inter- and intra-individual correspondence. To address this, we propose HippMetric, a skeletal representation (s-rep)-based framework for hippocampal substructural morphometry and point-wise correspondence across individuals and scans. HippMetric builds on the Axis-Referenced Morphometric Model (ARMM) and employs a deformable skeletal coordinate system aligned with hippocampal anatomy and function, providing a biologically grounded reference for correspondence. Our framework comprises two core modules: a skeletal-based coordinate system that respects the hippocampus' conserved longitudinal lamellar architecture, in which functional units (lamellae) are stacked perpendicular to the long-axis, enabling anatomically consistent localization across subjects and time; and individualized s-reps generated through surface reconstruction, deformation, and geometrically constrained spoke refinement, enforcing boundary adherence, orthogonality and non-intersection to produce mathematically valid skeletal geometry. Extensive experiments on two international cohorts demonstrate that HippMetric achieves higher accuracy, reliability, and correspondence stability compared to existing shape models.
Authors: Tiange Luo, Lajanugen Logeswaran, Jaekyeom Kim, Justin Johnson, Honglak Lee
Abstract: We introduce Image-LoRA, a lightweight parameter efficient fine-tuning (PEFT) recipe for transformer-based vision-language models (VLMs). Image-LoRA applies low-rank adaptation only to the value path of attention layers within the visual-token span, reducing adapter-only training FLOPs roughly in proportion to the visual-token fraction. We further adapt only a subset of attention heads, selected using head influence scores estimated with a rank-1 Image-LoRA, and stabilize per-layer updates via selection-size normalization. Across screen-centric grounding and referring benchmarks spanning text-heavy to image-heavy regimes, Image-LoRA matches or closely approaches standard LoRA accuracy while using fewer trainable parameters and lower adapter-only training FLOPs. The method also preserves the pure-text reasoning performance of VLMs before and after fine-tuning, as further shown on GSM8K.
Authors: Yunlong Liu, Shuyang Li, Pengyuan Liu, Yu Zhang, Rudi Stouffs
Abstract: Perception research is increasingly modelled using streetscapes, yet many approaches still rely on pixel features or object co-occurrence statistics, overlooking the explicit relations that shape human perception. This study proposes a three stage pipeline that transforms street view imagery (SVI) into structured representations for predicting six perceptual indicators. In the first stage, each image is parsed using an open-set Panoptic Scene Graph model (OpenPSG) to extract object predicate object triplets. In the second stage, compact scene-level embeddings are learned through a heterogeneous graph autoencoder (GraphMAE). In the third stage, a neural network predicts perception scores from these embeddings. We evaluate the proposed approach against image-only baselines in terms of accuracy, precision, and cross-city generalization. Results indicate that (i) our approach improves perception prediction accuracy by an average of 26% over baseline models, and (ii) maintains strong generalization performance in cross-city prediction tasks. Additionally, the structured representation clarifies which relational patterns contribute to lower perception scores in urban scenes, such as graffiti on wall and car parked on sidewalk. Overall, this study demonstrates that graph-based structure provides expressive, generalizable, and interpretable signals for modelling urban perception, advancing human-centric and context-aware urban analytics.
Authors: Meng Chu, Senqiao Yang, Haoxuan Che, Suiyun Zhang, Xichen Zhang, Shaozuo Yu, Haokun Gui, Zhefan Rao, Dandan Tu, Rui Liu, Jiaya Jia
Abstract: Generative models can now produce photorealistic imagery, yet they still struggle with the long, multi-goal prompts that professional designers issue. To expose this gap and better evaluate models' performance in real-world settings, we introduce Long Goal Bench (LGBench), a 2,000-task suite (1,000 T2I and 1,000 I2I) whose average instruction contains 18 to 22 tightly coupled goals spanning global layout, local object placement, typography, and logo fidelity. We find that even state-of-the-art models satisfy fewer than 72 percent of the goals and routinely miss localized edits, confirming the brittleness of current pipelines. To address this, we present VisionDirector, a training-free vision-language supervisor that (i) extracts structured goals from long instructions, (ii) dynamically decides between one-shot generation and staged edits, (iii) runs micro-grid sampling with semantic verification and rollback after every edit, and (iv) logs goal-level rewards. We further fine-tune the planner with Group Relative Policy Optimization, yielding shorter edit trajectories (3.1 versus 4.2 steps) and stronger alignment. VisionDirector achieves new state of the art on GenEval (plus 7 percent overall) and ImgEdit (plus 0.07 absolute) while producing consistent qualitative improvements on typography, multi-object scenes, and pose editing.
Authors: Xinyang Song, Libin Wang, Weining Wang, Zhiwei Li, Jianxin Sun, Dandan Zheng, Jingdong Chen, Qi Li, Zhenan Sun
Abstract: Recent image generation approaches often address subject, style, and structure-driven conditioning in isolation, leading to feature entanglement and limited task transferability. In this paper, we introduce 3SGen, a task-aware unified framework that performs all three conditioning modes within a single model. 3SGen employs an MLLM equipped with learnable semantic queries to align text-image semantics, complemented by a VAE branch that preserves fine-grained visual details. At its core, an Adaptive Task-specific Memory (ATM) module dynamically disentangles, stores, and retrieves condition-specific priors, such as identity for subjects, textures for styles, and spatial layouts for structures, via a lightweight gating mechanism along with several scalable memory items. This design mitigates inter-task interference and naturally scales to compositional inputs. In addition, we propose 3SGen-Bench, a unified image-driven generation benchmark with standardized metrics for evaluating cross-task fidelity and controllability. Extensive experiments on our proposed 3SGen-Bench and other public benchmarks demonstrate our superior performance across diverse image-driven generation tasks.
Authors: Ivan DeAndres-Tame, Chengwei Ye, Ruben Tolosana, Ruben Vera-Rodriguez, Shiqi Yu
Abstract: Generative AI (GenAI) models have revolutionized animation, enabling the synthesis of humans and motion patterns with remarkable visual fidelity. However, generating truly realistic human animation remains a formidable challenge, where even minor inconsistencies can make a subject appear unnatural. This limitation is particularly critical when AI-generated videos are evaluated for behavioral biometrics, where subtle motion cues that define identity are easily lost or distorted. The present study investigates whether state-of-the-art GenAI human animation models can preserve the subtle spatio-temporal details needed for person identification through gait biometrics. Specifically, we evaluate four different GenAI models across two primary evaluation tasks to assess their ability to i) restore gait patterns from reference videos under varying conditions of complexity, and ii) transfer these gait patterns to different visual identities. Our results show that while visual quality is mostly high, biometric fidelity remains low in tasks focusing on identification, suggesting that current GenAI models struggle to disentangle identity from motion. Furthermore, through an identity transfer task, we expose a fundamental flaw in appearance-based gait recognition: when texture is disentangled from motion, identification collapses, proving current GenAI models rely on visual attributes rather than temporal dynamics.
Authors: Kyungwon Cho, Hanbyul Joo
Abstract: Egocentric vision systems are becoming widely available, creating new opportunities for human-computer interaction. A core challenge is estimating the wearer's full-body motion from first-person videos, which is crucial for understanding human behavior. However, this task is difficult since most body parts are invisible from the egocentric view. Prior approaches mainly rely on head trajectories, leading to ambiguity, or assume continuously tracked hands, which is unrealistic for lightweight egocentric devices. In this work, we present HaMoS, the first hand-aware, sequence-level diffusion framework that directly conditions on both head trajectory and intermittently visible hand cues caused by field-of-view limitations and occlusions, as in real-world egocentric devices. To overcome the lack of datasets pairing diverse camera views with human motion, we introduce a novel augmentation method that models such real-world conditions. We also demonstrate that sequence-level contexts such as body shape and field-of-view are crucial for accurate motion reconstruction, and thus employ local attention to infer long sequences efficiently. Experiments on public benchmarks show that our method achieves state-of-the-art accuracy and temporal smoothness, demonstrating a practical step toward reliable in-the-wild egocentric 3D motion understanding.
Authors: Jun Li, Zikun Chen, Haibo Chen, Shuo Chen, Jian Yang
Abstract: Novel object synthesis by integrating distinct textual concepts from diverse categories remains a significant challenge in Text-to-Image (T2I) generation. Existing methods often suffer from insufficient concept mixing, lack of rigorous evaluation, and suboptimal outputs-manifesting as conceptual imbalance, superficial combinations, or mere juxtapositions. To address these limitations, we propose Reinforcement Mixing Learning (RMLer), a framework that formulates cross-category concept fusion as a reinforcement learning problem: mixed features serve as states, mixing strategies as actions, and visual outcomes as rewards. Specifically, we design an MLP-policy network to predict dynamic coefficients for blending cross-category text embeddings. We further introduce visual rewards based on (1) semantic similarity and (2) compositional balance between the fused object and its constituent concepts, optimizing the policy via proximal policy optimization. At inference, a selection strategy leverages these rewards to curate the highest-quality fused objects. Extensive experiments demonstrate RMLer's superiority in synthesizing coherent, high-fidelity objects from diverse categories, outperforming existing methods. Our work provides a robust framework for generating novel visual concepts, with promising applications in film, gaming, and design.
Authors: Xu Zhang, Junyao Ge, Yang Zheng, Kaitai Guo, Jimin Liang
Abstract: Large Vision-Language Models (LVLMs) hold great promise for advancing remote sensing (RS) analysis, yet existing reasoning segmentation frameworks couple linguistic reasoning and pixel prediction through end-to-end supervised fine-tuning, leading to weak geometric grounding and limited generalization across tasks. To address this, we developed Think2Seg-RS, a decoupled framework that trains an LVLM prompter to control a frozen Segment Anything Model (SAM) via structured geometric prompts. Through a mask-only reinforcement learning objective, the LVLM learns to translate abstract semantic reasoning into spatially grounded actions, achieving state-of-the-art performance on the EarthReason dataset. Remarkably, the learned prompting policy generalizes zero-shot to multiple referring segmentation benchmarks, exposing a distinct divide between semantic-level and instance-level grounding. We further found that compact segmenters outperform larger ones under semantic-level supervision, and that negative prompts are ineffective in heterogeneous aerial backgrounds. Together, these findings establish semantic-level reasoning segmentation as a new paradigm for geospatial understanding, opening the way toward unified, interpretable LVLM-driven Earth observation. Our code and model are available at https://github.com/Ricardo-XZ/Think2Seg-RS.
Authors: Hui Li, Jiayue Lyu, Fu-Yun Wang, Kaihui Cheng, Siyu Zhu, Jingdong Wang
Abstract: This paper studies the training-testing discrepancy (a.k.a. exposure bias) problem for improving the diffusion models. During training, the input of a prediction network at one training timestep is the corresponding ground-truth noisy data that is an interpolation of the noise and the data, and during testing, the input is the generated noisy data. We present a novel training approach, named MixFlow, for improving the performance. Our approach is motivated by the Slow Flow phenomenon: the ground-truth interpolation that is the nearest to the generated noisy data at a given sampling timestep is observed to correspond to a higher-noise timestep (termed slowed timestep), i.e., the corresponding ground-truth timestep is slower than the sampling timestep. MixFlow leverages the interpolations at the slowed timesteps, named slowed interpolation mixture, for post-training the prediction network for each training timestep. Experiments over class-conditional image generation (including SiT, REPA, and RAE) and text-to-image generation validate the effectiveness of our approach. Our approach MixFlow over the RAE models achieve strong generation results on ImageNet: 1.43 FID (without guidance) and 1.10 (with guidance) at 256 x 256, and 1.55 FID (without guidance) and 1.10 (with guidance) at 512 x 512.
Authors: Marica Muffoletto, Uxio Hermida, Charl\`ene Mauger, Avan Suinesiaputra, Yiyang Xu, Richard Burns, Lisa Pankewitz, Andrew D McCulloch, Steffen E Petersen, Daniel Rueckert, Alistair A Young
Abstract: Accurate reconstruction of cardiac anatomy from sparse clinical images remains a major challenge in patient-specific modeling. While neural implicit functions have previously been applied to this task, their application to mapping anatomical consistency across subjects has been limited. In this work, we introduce Neural Implicit Heart Coordinates (NIHCs), a standardized implicit coordinate system, based on universal ventricular coordinates, that provides a common anatomical reference frame for the human heart. Our method predicts NIHCs directly from a limited number of 2D segmentations (sparse acquisition) and subsequently decodes them into dense 3D segmentations and high-resolution meshes at arbitrary output resolution. Trained on a large dataset of 5,000 cardiac meshes, the model achieves high reconstruction accuracy on clinical contours, with mean Euclidean surface errors of 2.51$\pm$0.33 mm in a diseased cohort (n=4549) and 2.3$\pm$0.36 mm in a healthy cohort (n=5576). The NIHC representation enables anatomically coherent reconstruction even under severe slice sparsity and segmentation noise, faithfully recovering complex structures such as the valve planes. Compared with traditional pipelines, inference time is reduced from over 60 s to 5-15 s. These results demonstrate that NIHCs constitute a robust and efficient anatomical representation for patient-specific 3D cardiac reconstruction from minimal input data.
Authors: Moamal Fadhil Abdul (Department of Applied Mathematics and Computer Science, Technical University of Denmark, Richard Petersens Plads, Building 324, 2800 Kgs. Lyngby, Denmark), Jonas Bruun Hubrechts (Department of Applied Mathematics and Computer Science, Technical University of Denmark, Richard Petersens Plads, Building 324, 2800 Kgs. Lyngby, Denmark), Thomas Martini J{\o}rgensen (Department of Applied Mathematics and Computer Science, Technical University of Denmark, Richard Petersens Plads, Building 324, 2800 Kgs. Lyngby, Denmark), Emil Hovad (Department of Applied Mathematics and Computer Science, Technical University of Denmark, Richard Petersens Plads, Building 324, 2800 Kgs. Lyngby, Denmark)
Abstract: Automatically detecting and classifying strokes in table tennis video can streamline training workflows, enrich broadcast overlays, and enable fine-grained performance analytics. For this to be possible, annotated video data of table tennis is needed. We extend the public OpenTTGames dataset with highly detailed, frame-accurate shot type annotations (forehand, backhand with subtypes), player posture labels (body lean and leg stance), and rally outcome tags at point end. OpenTTGames is a set of recordings from the side of the table with official labels for bounces, when the ball is above the net, or hitting the net. The dataset already contains ball coordinates near events, which are either "bounce", "net", or "empty_event" in the original OpenTTGames dataset, and semantic masks (humans, table, scoreboard). Our extension adds the types of stroke to the events and a per-player taxonomy so models can move beyond event spotting toward tactical understanding (e.g., whether a stroke is likely to win the point or set up an advantage). We provide a compact coding scheme and code-assisted labeling procedure to support reproducible annotations and baselines for fine-grained stroke understanding in racket sports. This fills a practical gap in the community, where many prior video resources are either not publicly released or carry restrictive/unclear licenses that hinder reuse and benchmarking. Our annotations are released under the same CC BY-NC-SA 4.0 license as OpenTTGames, allowing free non-commercial use, modification, and redistribution, with appropriate attribution.
Authors: Yueting Zhu, Yuehao Song, Shuai Zhang, Wenyu Liu, Xinggang Wang
Abstract: Whole Slide Images (WSIs) are typically analyzed using multiple instance learning (MIL) methods. However, the scale and heterogeneity of WSIs generate highly redundant and dispersed information, making it difficult to identify and integrate discriminative signals. Existing MIL methods either fail to discard uninformative cues effectively or have limited ability to consolidate relevant features from multiple patches, which restricts their performance on large and heterogeneous WSIs. To address this issue, we propose DeltaMIL, a novel MIL framework that explicitly selects semantically relevant regions and integrates the discriminative information from WSIs. Our method leverages the gated delta rule to efficiently filter and integrate information through a block combining forgetting and memory mechanisms. The delta mechanism dynamically updates the memory by removing old values and inserting new ones according to their correlation with the current patch. The gating mechanism further enables rapid forgetting of irrelevant signals. Additionally, DeltaMIL integrates a complementary local pattern mixing mechanism to retain fine-grained pathological locality. Our design enhances the extraction of meaningful cues and suppresses redundant or noisy information, which improves the model's robustness and discriminative power. Experiments demonstrate that DeltaMIL achieves state-of-the-art performance. Specifically, for survival prediction, DeltaMIL improves performance by 3.69\% using ResNet-50 features and 2.36\% using UNI features. For slide-level classification, it increases accuracy by 3.09\% with ResNet-50 features and 3.75\% with UNI features. These results demonstrate the strong and consistent performance of DeltaMIL across diverse WSI tasks.
Authors: Siyuan Mei, Yan Xia, Fuxin Fan
Abstract: The synthesis of computed tomography (CT) from magnetic resonance imaging (MRI) and cone-beam CT (CBCT) plays a critical role in clinical treatment planning by enabling accurate anatomical representation in adaptive radiotherapy. In this work, we propose GANeXt, a 3D patch-based, fully ConvNeXt-powered generative adversarial network for unified CT synthesis across different modalities and anatomical regions. Specifically, GANeXt employs an efficient U-shaped generator constructed from stacked 3D ConvNeXt blocks with compact convolution kernels, while the discriminator adopts a conditional PatchGAN. To improve synthesis quality, we incorporate a combination of loss functions, including mean absolute error (MAE), perceptual loss, segmentation-based masked MAE, and adversarial loss and a combination of Dice loss and cross-entropy for multi-head segmentation discriminator. For both tasks, training is performed with a batch size of 8 using two separate AdamW optimizers for the generator and discriminator, each equipped with a warmup and cosine decay scheduler, with learning rates of $5\times10^{-4}$ and $1\times10^{-3}$, respectively. Data preprocessing includes deformable registration, foreground cropping, percentile normalization for the input modality, and linear normalization of the CT to the range $[-1024, 1000]$. Data augmentation involves random zooming within $(0.8, 1.3)$ (for MRI-to-CT only), fixed-size cropping to $32\times160\times192$ for MRI-to-CT and $32\times128\times128$ for CBCT-to-CT, and random flipping. During inference, we apply a sliding-window approach with $0.8$ overlap and average folding to reconstruct the full-size sCT, followed by inversion of the CT normalization. After joint training on all regions without any fine-tuning, the final models are selected at the end of 3000 epochs for MRI-to-CT and 1000 epochs for CBCT-to-CT using the full training dataset.
Authors: Zhenyang Huang, Xiao Yu, Yi Zhang, Decheng Wang, Hang Ruan
Abstract: Remote sensing image change detection is one of the fundamental tasks in remote sensing intelligent interpretation. Its core objective is to identify changes within change regions of interest (CRoI). Current multimodal large models encode rich human semantic knowledge, which is utilized for guidance in tasks such as remote sensing change detection. However, existing methods that use semantic guidance for detecting users' CRoI overly rely on explicit textual descriptions of CRoI, leading to the problem of near-complete performance failure when presented with implicit CRoI textual descriptions. This paper proposes a multimodal reasoning change detection model named ReasonCD, capable of mining users' implicit task intent. The model leverages the powerful reasoning capabilities of pre-trained large language models to mine users' implicit task intents and subsequently obtains different change detection results based on these intents. Experiments on public datasets demonstrate that the model achieves excellent change detection performance, with an F1 score of 92.1\% on the BCDD dataset. Furthermore, to validate its superior reasoning functionality, this paper annotates a subset of reasoning data based on the SECOND dataset. Experimental results show that the model not only excels at basic reasoning-based change detection tasks but can also explain the reasoning process to aid human decision-making.
Authors: Zhongwei Chen, Hai-Jun Rong, Zhao-Xu Yang, Guoqi Li
Abstract: Traditional drone-view geo-localization (DVGL) methods based on artificial neural networks (ANNs) have achieved remarkable performance. However, ANNs rely on dense computation, which results in high power consumption. In contrast, spiking neural networks (SNNs), which benefit from spike-driven computation, inherently provide low power consumption. Regrettably, the potential of SNNs for DVGL has yet to be thoroughly investigated. Meanwhile, the inherent sparsity of spike-driven computation for representation learning scenarios also results in loss of critical information and difficulties in learning long-range dependencies when aligning heterogeneous visual data sources. To address these, we propose SpikeViMFormer, the first SNN framework designed for DVGL. In this framework, a lightweight spike-driven transformer backbone is adopted to extract coarse-grained features. To mitigate the loss of critical information, the spike-driven selective attention (SSA) block is designed, which uses a spike-driven gating mechanism to achieve selective feature enhancement and highlight discriminative regions. Furthermore, a spike-driven hybrid state space (SHS) block is introduced to learn long-range dependencies using a hybrid state space. Moreover, only the backbone is utilized during the inference stage to reduce computational cost. To ensure backbone effectiveness, a novel hierarchical re-ranking alignment learning (HRAL) strategy is proposed. It refines features via neighborhood re-ranking and maintains cross-batch consistency to directly optimize the backbone. Experimental results demonstrate that SpikeViMFormer outperforms state-of-the-art SNNs. Compared with advanced ANNs, it also achieves competitive performance.Our code is available at https://github.com/ISChenawei/SpikeViMFormer
Authors: Yueyao Chen, Kai-Ni Wang, Dario Tayupo, Arnaud Huaulm'e, Krystel Nyangoh Timoh, Pierre Jannin, Qi Dou
Abstract: Purpose: Surgical workflow recognition enables context-aware assistance and skill assessment in computer-assisted interventions. Despite recent advances, current methods suffer from two critical challenges: prediction jitter across consecutive frames and poor discrimination of ambiguous phases. This paper aims to develop a stable framework by selectively propagating reliable historical information and explicitly modeling uncertainty for hard sample enhancement. Methods: We propose a dual-pathway framework DSTED with Reliable Memory Propagation (RMP) and Uncertainty-Aware Prototype Retrieval (UPR). RMP maintains temporal coherence by filtering and fusing high-confidence historical features through multi-criteria reliability assessment. UPR constructs learnable class-specific prototypes from high-uncertainty samples and performs adaptive prototype matching to refine ambiguous frame representations. Finally, a confidence-driven gate dynamically balances both pathways based on prediction certainty. Results: Our method achieves state-of-the-art performance on AutoLaparo-hysterectomy with 84.36% accuracy and 65.51% F1-score, surpassing the second-best method by 3.51% and 4.88% respectively. Ablations reveal complementary gains from RMP (2.19%) and UPR (1.93%), with synergistic effects when combined. Extensive analysis confirms substantial reduction in temporal jitter and marked improvement on challenging phase transitions. Conclusion: Our dual-pathway design introduces a novel paradigm for stable workflow recognition, demonstrating that decoupling the modeling of temporal consistency and phase ambiguity yields superior performance and clinical applicability.
Authors: Xiaoming Zhang, Chunli Li, Jiacheng Hao, Yuan Gao, Danyang Tu, Jianyi Qiao, Xiaoli Yin, Le Lu, Ling Zhang, Ke Yan, Yang Hou, Yu Shi
Abstract: Esophageal varices (EV) represent a critical complication of portal hypertension, affecting approximately 60% of cirrhosis patients with a significant bleeding risk of ~30%. While traditionally diagnosed through invasive endoscopy, non-contrast computed tomography (NCCT) presents a potential non-invasive alternative that has yet to be fully utilized in clinical practice. We present Multi-Organ-COhesion Network++ (MOON++), a novel multimodal framework that enhances EV assessment through comprehensive analysis of NCCT scans. Inspired by clinical evidence correlating organ volumetric relationships with liver disease severity, MOON++ synthesizes imaging characteristics of the esophagus, liver, and spleen through multimodal learning. We evaluated our approach using 1,631 patients, those with endoscopically confirmed EV were classified into four severity grades. Validation in 239 patient cases and independent testing in 289 cases demonstrate superior performance compared to conventional single organ methods, achieving an AUC of 0.894 versus 0.803 for the severe grade EV classification (G3 versus
Authors: Yi Xin, Siqi Luo, Qi Qin, Haoxing Chen, Kaiwen Zhu, Zhiwei Zhang, Yangfan He, Rongchao Zhang, Jinbin Bai, Shuo Cao, Bin Fu, Junjun He, Yihao Liu, Yuewen Cao, Xiaohong Liu
Abstract: Diffusion Multi-modal Large Language Models (dMLLMs) have recently emerged as a novel architecture unifying image generation and understanding. However, developing effective and efficient Test-Time Scaling (TTS) methods to unlock their full generative potential remains an underexplored challenge. To address this, we propose dMLLM-TTS, a novel framework operating on two complementary scaling axes: (1) trajectory exploration scaling to enhance the diversity of generated hypotheses, and (2) iterative refinement scaling for stable generation. Conventional TTS approaches typically perform linear search across these two dimensions, incurring substantial computational costs of O(NT) and requiring an external verifier for best-of-N selection. To overcome these limitations, we propose two innovations. First, we design an efficient hierarchical search algorithm with O(N+T) complexity that adaptively expands and prunes sampling trajectories. Second, we introduce a self-verified feedback mechanism that leverages the dMLLMs' intrinsic image understanding capabilities to assess text-image alignment, eliminating the need for external verifier. Extensive experiments on the GenEval benchmark across three representative dMLLMs (e.g., Lumina-DiMOO, MMaDA, Muddit) show that our framework substantially improves generation quality while achieving up to 6x greater efficiency than linear search. Project page: https://github.com/Alpha-VLLM/Lumina-DiMOO.
Authors: Fei Ge, Ying Huang, Jie Liu, Guixuan Zhang, Zhi Zeng, Shuwu Zhang, Hu Guan
Abstract: Existing deep image watermarking methods follow a fixed embedding-distortion-extraction pipeline, where the embedder and extractor are weakly coupled through a final loss and optimized in isolation. This design lacks explicit collaboration, leaving no structured mechanism for the embedder to incorporate decoding-aware cues or for the extractor to guide embedding during training. To address this architectural limitation, we rethink deep image watermarking by reformulating embedding and extraction as explicitly collaborative components. To realize this reformulation, we introduce a Collaborative Interaction Mechanism (CIM) that establishes direct, bidirectional communication between the embedder and extractor, enabling a mutual-teacher training paradigm and coordinated optimization. Built upon this explicitly collaborative architecture, we further propose an Adaptive Feature Modulation Module (AFMM) to support effective interaction. AFMM enables content-aware feature regulation by decoupling modulation structure and strength, guiding watermark embedding toward stable image features while suppressing host interference during extraction. Under CIM, the AFMMs on both sides form a closed-loop collaboration that aligns embedding behavior with extraction objectives. This architecture-level redesign changes how robustness is learned in watermarking systems. Rather than relying on exhaustive distortion simulation, robustness emerges from coordinated representation learning between embedding and extraction. Experiments on real-world and AI-generated datasets demonstrate that the proposed method consistently outperforms state-of-the-art approaches in watermark extraction accuracy while maintaining high perceptual quality, showing strong robustness and generalization.
Authors: Evelyn Zhang, Fufu Yu, Aoqi Wu, Zichen Wen, Ke Yan, Shouhong Ding, Biqing Qi, Linfeng Zhang
Abstract: Processing long visual token sequences poses a significant computational burden on Multimodal Large Language Models (MLLMs). While token pruning offers a path to acceleration, we find that current methods, while adequate for general understanding, catastrophically fail on fine-grained localization tasks. We attribute this failure to the inherent flaws of the two prevailing strategies: importance-based methods suffer from a strong positional bias, an inherent model artifact that distracts from semantic content, while diversity-based methods exhibit structural blindness, disregarding the user's prompt and spatial redundancy. To address this, we introduce D2Pruner, a framework that rectifies these issues by uniquely combining debiased importance with a structural pruning mechanism. Our method first secures a core set of the most critical tokens as pivots based on a debiased attention score. It then performs a Maximal Independent Set (MIS) selection on the remaining tokens, which are modeled on a hybrid graph where edges signify spatial proximity and semantic similarity. This process iteratively preserves the most important and available token while removing its neighbors, ensuring that the supplementary tokens are chosen to maximize importance and diversity. Extensive experiments demonstrate that D2Pruner has exceptional efficiency and fidelity. Applied to LLaVA-1.5-7B for general understanding tasks, it reduces FLOPs by 74.2\% while retaining 99.2\% of its original performance. Furthermore, in challenging localization benchmarks with InternVL-2.5-8B, it maintains 85.7\% performance at a 90\% token reduction rate, marking a significant advancement with up to 63. 53\% improvement over existing methods.
Authors: Nitin Kumar Singh, Arie Rachmad Syulistyo, Yuichiro Tanaka, Hakaru Tamukoh
Abstract: Sign language recognition (SLR) facilitates communication between deaf and hearing communities. Deep learning based SLR models are commonly used but require extensive computational resources, making them unsuitable for deployment on edge devices. To address these limitations, we propose a lightweight SLR system that combines parallel bidirectional reservoir computing (PBRC) with MediaPipe. MediaPipe enables real-time hand tracking and precise extraction of hand joint coordinates, which serve as input features for the PBRC architecture. The proposed PBRC architecture consists of two echo state network (ESN) based bidirectional reservoir computing (BRC) modules arranged in parallel to capture temporal dependencies, thereby creating a rich feature representation for classification. We trained our PBRC-based SLR system on the Word-Level American Sign Language (WLASL) video dataset, achieving top-1, top-5, and top-10 accuracies of 60.85%, 85.86%, and 91.74%, respectively. Training time was significantly reduced to 18.67 seconds due to the intrinsic properties of reservoir computing, compared to over 55 minutes for deep learning based methods such as Bi-GRU. This approach offers a lightweight, cost-effective solution for real-time SLR on edge devices.
Authors: Guoli Jia, Junyao Hu, Xinwei Long, Kai Tian, Kaiyan Zhang, KaiKai Zhao, Ning Ding, Bowen Zhou
Abstract: Image generation based on diffusion models has demonstrated impressive capability, motivating exploration into diverse and specialized applications. Owing to the importance of emotion in advertising, emotion-oriented image generation has attracted increasing attention. However, current emotion-oriented methods suffer from an affective shortcut, where emotions are approximated to semantics. As evidenced by two decades of research, emotion is not equivalent to semantics. To this end, we propose Emotion-Director, a cross-modal collaboration framework consisting of two modules. First, we propose a cross-Modal Collaborative diffusion model, abbreviated as MC-Diffusion. MC-Diffusion integrates visual prompts with textual prompts for guidance, enabling the generation of emotion-oriented images beyond semantics. Further, we improve the DPO optimization by a negative visual prompt, enhancing the model's sensitivity to different emotions under the same semantics. Second, we propose MC-Agent, a cross-Modal Collaborative Agent system that rewrites textual prompts to express the intended emotions. To avoid template-like rewrites, MC-Agent employs multi-agents to simulate human subjectivity toward emotions, and adopts a chain-of-concept workflow that improves the visual expressiveness of the rewritten prompts. Extensive qualitative and quantitative experiments demonstrate the superiority of Emotion-Director in emotion-oriented image generation.
Authors: Shaochen Bi, Yuting He, Weiming Wang, Hao Chen
Abstract: Combinatorial explosion problem caused by dual inputs presents a critical challenge in Deformable Medical Image Registration (DMIR). Since DMIR processes two images simultaneously as input, the combination relationships between features has grown exponentially, ultimately the model considers more interfering features during the feature modeling process. Introducing dynamics in the receptive fields and weights of the network enable the model to eliminate the interfering features combination and model the potential feature combination relationships. In this paper, we propose the Dynamic Stream Network (DySNet), which enables the receptive fields and weights to be dynamically adjusted. This ultimately enables the model to ignore interfering feature combinations and model the potential feature relationships. With two key innovations: 1) Adaptive Stream Basin (AdSB) module dynamically adjusts the shape of the receptive field, thereby enabling the model to focus on the feature relationships with greater correlation. 2) Dynamic Stream Attention (DySA) mechanism generates dynamic weights to search for more valuable feature relationships. Extensive experiments have shown that DySNet consistently outperforms the most advanced DMIR methods, highlighting its outstanding generalization ability. Our code will be released on the website: https://github.com/ShaochenBi/DySNet.
Authors: Georgios Voulgaris
Abstract: Modern deep learning models operating on multi-modal visual signals often rely on inductive biases that are poorly aligned with the physical processes governing signal formation, leading to brittle performance under cross-spectral and real-world conditions. In particular, approaches that prioritise direct thermal cues struggle to capture indirect yet persistent environmental alterations induced by sustained heat emissions. This work introduces a physics-aware representation learning framework that leverages multi-spectral information to model stable signatures of long-term physical processes. Specifically, a geological Short Wave Infrared (SWIR) ratio sensitive to soil property changes is integrated with Thermal Infrared (TIR) data through an intermediate fusion architecture, instantiated as FusionNet. The proposed backbone embeds trainable differential signal-processing priors within convolutional layers, combines mixed pooling strategies, and employs wider receptive fields to enhance robustness across spectral modalities. Systematic ablations show that each architectural component contributes to performance gains, with DGCNN achieving 88.7% accuracy on the SWIR ratio and FusionNet reaching 90.6%, outperforming state-of-the-art baselines across five spectral configurations. Transfer learning experiments further show that ImageNet pretraining degrades TIR performance, highlighting the importance of modality-aware training for cross-spectral learning. Evaluated on real-world data, the results demonstrate that combining physics-aware feature selection with principled deep learning architectures yields robust and generalisable representations, illustrating how first-principles signal modelling can improve multi-spectral learning under challenging conditions.
Authors: Ziyang Song, Zelin Zang, Zuyao Chen, Xusheng Liang, Dong Yi, Jinlin Wu, Hongbin Liu, Jiebo Luo
Abstract: Multimodal Large Language Models (MLLMs) have achieved impressive progress in natural image reasoning, yet their potential in medical imaging remains underexplored, especially in clinical anatomical surgical images. Anatomy understanding tasks demand precise understanding and clinically coherent answers, which are difficult to achieve due to the complexity of medical data and the scarcity of high-quality expert annotations. These challenges limit the effectiveness of conventional Supervised Fine-Tuning (SFT) strategies. While recent work has demonstrated that Group Relative Policy Optimization (GRPO) can enhance reasoning in MLLMs without relying on large amounts of data, we find two weaknesses that hinder GRPO's reasoning performance in anatomy recognition: 1) knowledge cannot be effectively shared between different anatomical structures, resulting in uneven information gain and preventing the model from converging, and 2) the model quickly converges to a single reasoning path, suppressing the exploration of diverse strategies. To overcome these challenges, we propose two novel methods. First, we implement a progressive learning strategy called Anatomical Similarity Curriculum Learning by controlling question difficulty via the similarity of answer choices, enabling the model to master complex problems incrementally. Second, we utilize question augmentation referred to as Group Diversity Question Augmentation to expand the model's search space for difficult queries, mitigating the tendency to produce uniform responses. Comprehensive experiments on the SGG-VQA and OmniMedVQA benchmarks show our method achieves a significant improvement across the two benchmarks, demonstrating its effectiveness in enhancing the medical reasoning capabilities of MLLMs. The code can be found in https://github.com/tomato996/Anatomy-R1
Authors: Zhuo He, Yingdong Ru, Qianying Liu, Paul Henderson, Nicolas Pugeault
Abstract: Recent advances in neural rendering have achieved impressive results on photorealistic shading and relighting, by using a multilayer perceptron (MLP) as a regression model to learn the rendering equation from a real-world dataset. Such methods show promise for photorealistically relighting real-world objects, which is difficult to classical rendering, as there is no easy-obtained material ground truth. However, significant challenges still remain the dense connections in MLPs result in a large number of parameters, which requires high computation resources, complicating the training, and reducing performance during rendering. Data driven approaches require large amounts of training data for generalization; unbalanced data might bias the model to ignore the unusual illumination conditions, e.g. dark scenes. This paper introduces pbnds+: a novel physics-based neural deferred shading pipeline utilizing convolution neural networks to decrease the parameters and improve the performance in shading and relighting tasks; Energy regularization is also proposed to restrict the model reflection during dark illumination. Extensive experiments demonstrate that our approach outperforms classical baselines, a state-of-the-art neural shading model, and a diffusion-based method.
Authors: Marc Peral, Guillem Capellera, Luis Ferraz, Antonio Rubio, Antonio Agudo
Abstract: In this work we propose a multi-modal architecture for analyzing soccer scenes from tactical camera footage, with a focus on three core tasks: ball trajectory inference, ball state classification, and ball possessor identification. To this end, our solution integrates three distinct input modalities (player trajectories, player types and image crops of individual players) into a unified framework that processes spatial and temporal dynamics using a cascade of sociotemporal transformer blocks. Unlike prior methods, which rely heavily on accurate ball tracking or handcrafted heuristics, our approach infers the ball trajectory without direct access to its past or future positions, and robustly identifies the ball state and ball possessor under noisy or occluded conditions from real top league matches. We also introduce CropDrop, a modality-specific masking pre-training strategy that prevents over-reliance on image features and encourages the model to rely on cross-modal patterns during pre-training. We show the effectiveness of our approach on a large-scale dataset providing substantial improvements over state-of-the-art baselines in all tasks. Our results highlight the benefits of combining structured and visual cues in a transformer-based architecture, and the importance of realistic masking strategies in multi-modal learning.
Authors: Chi Zhang, Braedon Gunn, Andrew M. Read-Fuller
Abstract: Poor adaptation of orbital implants remains a major contributor to postoperative complications and revision surgery. Although preformed orbital plates are widely used to reduce cost and operative time compared with customized implants, surgeons currently lack publicly available tools and standardized metrics to quantitatively compare plate fit across vendors, sizes, and patient anatomy. We developed SlicerOrbitSurgerySim, an open-source extension for the 3D Slicer platform that enables interactive virtual registration, evaluation, and comparison of multiple preformed orbital plates in a patient-specific virtual planning environment. The software generates reproducible quantitative plate-to-orbit distance metrics and visualization tools that support both patient-specific planning and population-level statistical analysis of plate adaptability. By facilitating objective comparison of implant designs and placement strategies, this tool aims to improve preoperative decision-making, reduce intraoperative plate modification, and promote collaborative research and surgical education. Pilot studies, sample datasets, and detailed tutorials are provided to support testing, transparency, and reproducibility.
Authors: Moritz B\"ohle, Am\'elie Royer, Juliette Marrie, Edouard Grave, Patrick P\'erez
Abstract: Vision-language models (VLMs) are commonly trained by inserting image tokens from a pretrained vision encoder into the textual stream of a language model. This allows text and image information to fully attend to one another within the model, but becomes extremely costly for high-resolution images, long conversations, or streaming videos, both in memory and compute. VLMs leveraging cross-attention are an efficient alternative to token insertion but exhibit a clear performance gap, in particular on tasks involving fine-grained visual details. We find that a key to improving such models is to also enable local text-to-text interaction in the dedicated cross-attention layers. Building on this, we propose CASA, Cross-Attention via Self-Attention, a simple and efficient paradigm which substantially reduces the gap with full token insertion on common image understanding benchmarks, while enjoying the same scalability as cross-attention models when applied to long-context multimodal tasks such as streaming video captioning. For samples and code, please see our project page at https://kyutai.org/casa .
URLs: https://kyutai.org/casa
Authors: Kaiwen Zhang, Liming Jiang, Angtian Wang, Jacob Zhiyuan Fang, Tiancheng Zhi, Qing Yan, Hao Kang, Xin Lu, Xingang Pan
Abstract: Visual storytelling requires generating multi-shot videos with cinematic quality and long-range consistency. Inspired by human memory, we propose StoryMem, a paradigm that reformulates long-form video storytelling as iterative shot synthesis conditioned on explicit visual memory, transforming pre-trained single-shot video diffusion models into multi-shot storytellers. This is achieved by a novel Memory-to-Video (M2V) design, which maintains a compact and dynamically updated memory bank of keyframes from historical generated shots. The stored memory is then injected into single-shot video diffusion models via latent concatenation and negative RoPE shifts with only LoRA fine-tuning. A semantic keyframe selection strategy, together with aesthetic preference filtering, further ensures informative and stable memory throughout generation. Moreover, the proposed framework naturally accommodates smooth shot transitions and customized story generation applications. To facilitate evaluation, we introduce ST-Bench, a diverse benchmark for multi-shot video storytelling. Extensive experiments demonstrate that StoryMem achieves superior cross-shot consistency over previous methods while preserving high aesthetic quality and prompt adherence, marking a significant step toward coherent minute-long video storytelling.
Authors: Ziqiao Peng, Yi Chen, Yifeng Ma, Guozhen Zhang, Zhiyao Sun, Zixiang Zhou, Youliang Zhang, Zhengguang Zhou, Zhaoxin Fan, Hongyan Liu, Yuan Zhou, Qinglin Lu, Jun He
Abstract: Despite significant advances in talking avatar generation, existing methods face critical challenges: insufficient text-following capability for diverse actions, lack of temporal alignment between actions and audio content, and dependency on additional control signals such as pose skeletons. We present ActAvatar, a framework that achieves phase-level precision in action control through textual guidance by capturing both action semantics and temporal context. Our approach introduces three core innovations: (1) Phase-Aware Cross-Attention (PACA), which decomposes prompts into a global base block and temporally-anchored phase blocks, enabling the model to concentrate on phase-relevant tokens for precise temporal-semantic alignment; (2) Progressive Audio-Visual Alignment, which aligns modality influence with the hierarchical feature learning process-early layers prioritize text for establishing action structure while deeper layers emphasize audio for refining lip movements, preventing modality interference; (3) A two-stage training strategy that first establishes robust audio-visual correspondence on diverse data, then injects action control through fine-tuning on structured annotations, maintaining both audio-visual alignment and the model's text-following capabilities. Extensive experiments demonstrate that ActAvatar significantly outperforms state-of-the-art methods in both action control and visual quality.
Authors: Antonia Alomar, Mireia Masias, Marius George Linguraru, Federico M. Sukno, Gemma Piella
Abstract: Early detection of developmental disorders can be aided by analyzing infant craniofacial morphology, but modeling infant faces is challenging due to limited data and frequent spontaneous expressions. We introduce BabyFlow, a generative AI model that disentangles facial identity and expression, enabling independent control over both. Using normalizing flows, BabyFlow learns flexible, probabilistic representations that capture the complex, non-linear variability of expressive infant faces without restrictive linear assumptions. To address scarce and uncontrolled expressive data, we perform cross-age expression transfer, adapting expressions from adult 3D scans to enrich infant datasets with realistic and systematic expressive variants. As a result, BabyFlow improves 3D reconstruction accuracy, particularly in highly expressive regions such as the mouth, eyes, and nose, and supports synthesis and modification of infant expressions while preserving identity. Additionally, by integrating with diffusion models, BabyFlow generates high-fidelity 2D infant images with consistent 3D geometry, providing powerful tools for data augmentation and early facial analysis.
Authors: Marta Hasny, Laura Daza, Keno Bressem, Maxime Di Folco, Julia Schnabel
Abstract: Large-scale medical biobanks provide imaging data complemented by extensive tabular information, such as demographics or clinical measurements. However, this abundance of tabular attributes does not reflect real-world datasets, where only a subset of attributes may be available. This discrepancy calls for methods that can leverage all the tabular data during training while remaining robust to missing values at inference. To address this challenge, we propose RoVTL (Robust Vision-Tabular Learning), a framework designed to handle any level of tabular data availability, from 0% to 100%. RoVTL comprises two key stages: contrastive pretraining, where we introduce tabular attribute missingness as data augmentation to promote robustness, and downstream task tuning using a gated cross-attention module for multimodal fusion. During fine-tuning, we employ a novel Tabular More vs. Fewer loss that ranks performance based on the amount of available tabular data. Combined with disentangled gradient learning, this enables consistent performance across all tabular data completeness scenarios. We evaluate RoVTL on cardiac MRI scans from the UK Biobank, demonstrating superior robustness to missing tabular data compared to prior methods. Furthermore, RoVTL successfully generalizes to an external cardiac MRI dataset for multimodal disease classification, and extends to the natural images domain, achieving robust performance on a car advertisements dataset. The code is available at https://github.com/marteczkah/RoVTL.
Authors: Artemis Panagopoulou, Aveek Purohit, Achin Kulshrestha, Soroosh Yazdani, Mohit Goyal
Abstract: While Multimodal Large Language Models have achieved human-like performance on many visual and textual reasoning tasks, their proficiency in fine-grained spatial understanding, such as route tracing on maps remains limited. Unlike humans, who can quickly learn to parse and navigate maps, current models often fail to respect fundamental path constraints, in part due to the prohibitive cost and difficulty of collecting large-scale, pixel-accurate path annotations. To address this, we introduce a scalable synthetic data generation pipeline that leverages synthetic map images and pixel-level parsing to automatically produce precise annotations for this challenging task. Using this pipeline, we construct a fine-tuning dataset of 23k path samples across 4k maps, enabling models to acquire more human-like spatial capabilities. Using this dataset, we fine-tune both open-source and proprietary MLLMs. Results on MapBench show that finetuning substantially improves robustness, raising success rates by up to 6.4 points, while also reducing path-tracing error (NDTW). These gains highlight that fine-grained spatial reasoning, absent in pretrained models, can be explicitly taught with synthetic supervision.
Authors: Da Tan, Michael Beck, Christopher P. Bidinosti, Robert H. Gulden, Christopher J. Henry
Abstract: The success of agricultural artificial intelligence depends heavily on large, diverse, and high-quality plant image datasets, yet collecting such data in real field conditions is costly, labor intensive, and seasonally constrained. This paper investigates diffusion-based generative modeling to address these challenges through plant image synthesis, indoor-to-outdoor translation, and expert preference aligned fine tuning. First, a Stable Diffusion model is fine tuned on captioned indoor and outdoor plant imagery to generate realistic, text conditioned images of canola and soybean. Evaluation using Inception Score, Frechet Inception Distance, and downstream phenotype classification shows that synthetic images effectively augment training data and improve accuracy. Second, we bridge the gap between high resolution indoor datasets and limited outdoor imagery using DreamBooth-based text inversion and image guided diffusion, generating translated images that enhance weed detection and classification with YOLOv8. Finally, a preference guided fine tuning framework trains a reward model on expert scores and applies reward weighted updates to produce more stable and expert aligned outputs. Together, these components demonstrate a practical pathway toward data efficient generative pipelines for agricultural AI.
Authors: Arnold Caleb Asiimwe, Carl Vondrick
Abstract: We reinterpret 4D Gaussian Splatting as a continuous-time dynamical system, where scene motion arises from integrating a learned neural dynamical field rather than applying per-frame deformations. This formulation, which we call EvoGS, treats the Gaussian representation as an evolving physical system whose state evolves continuously under a learned motion law. This unlocks capabilities absent in deformation-based approaches:(1) sample-efficient learning from sparse temporal supervision by modeling the underlying motion law; (2) temporal extrapolation enabling forward and backward prediction beyond observed time ranges; and (3) compositional dynamics that allow localized dynamics injection for controllable scene synthesis. Experiments on dynamic scene benchmarks show that EvoGS achieves better motion coherence and temporal consistency compared to deformation-field baselines while maintaining real-time rendering
Authors: Luchao Qi, Jiaye Wu, Jun Myeong Choi, Cary Phillips, Roni Sengupta, Dan B Goldman
Abstract: In professional video compositing workflows, artists must manually create environmental interactions-such as shadows, reflections, dust, and splashes-between foreground subjects and background layers. Existing video generative models struggle to preserve the input video while adding such effects, and current video inpainting methods either require costly per-frame masks or yield implausible results. We introduce augmented compositing, a new task that synthesizes realistic, semi-transparent environmental effects conditioned on text prompts and input video layers, while preserving the original scene. To address this task, we present Over++, a video effect generation framework that makes no assumptions about camera pose, scene stationarity, or depth supervision. We construct a paired effect dataset tailored for this task and introduce an unpaired augmentation strategy that preserves text-driven editability. Our method also supports optional mask control and keyframe guidance without requiring dense annotations. Despite training on limited data, Over++ produces diverse and realistic environmental effects and outperforms existing baselines in both effect generation and scene preservation.
Authors: Argha Kamal Samanta, Harshika Goyal, Vasudha Joshi, Tushar Mungle, Pabitra Mitra
Abstract: Diabetic retinopathy (DR) is a leading cause of preventable blindness worldwide, demanding accurate automated diagnostic systems. While general-domain vision-language models like Contrastive Language-Image Pre-Training (CLIP) perform well on natural image tasks, they struggle in medical domain applications, particularly in cross-modal retrieval for ophthalmological images. We propose a novel knowledge-enhanced joint embedding framework that integrates retinal fundus images, clinical text, and structured patient data through a multimodal transformer architecture to address the critical gap in medical image-text alignment. Our approach employs separate encoders for each modality: a Vision Transformer (ViT-B/16) for retinal images, Bio-ClinicalBERT for clinical narratives, and a multilayer perceptron for structured demographic and clinical features. These modalities are fused through a joint transformer with modality-specific embeddings, trained using multiple objectives including contrastive losses between modality pairs, reconstruction losses for images and text, and classification losses for DR severity grading according to ICDR and SDRG schemes. Experimental results on the Brazilian Multilabel Ophthalmological Dataset (BRSET) demonstrate significant improvements over baseline models. Our framework achieves near-perfect text-to-image retrieval performance with Recall@1 of 99.94% compared to fine-tuned CLIP's 1.29%, while maintaining state-of-the-art classification accuracy of 97.05% for SDRG and 97.97% for ICDR. Furthermore, zero-shot evaluation on the unseen DeepEyeNet dataset validates strong generalizability with 93.95% Recall@1 versus 0.22% for fine-tuned CLIP. These results demonstrate that our multimodal training approach effectively captures cross-modal relationships in the medical domain, establishing both superior retrieval capabilities and robust diagnostic performance.
Authors: Mojtaba Safari, Shansong Wang, Vanessa L Wildman, Mingzhe Hu, Zach Eidex, Chih-Wei Chang, Erik H Middlebrooks, Richard L. J Qiu, Pretesh Patel, Ashesh B. Jania, Hui Mao, Zhen Tian, Xiaofeng Yang
Abstract: Background: High-resolution MRI is critical for diagnosis, but long acquisition times limit clinical use. Super-resolution (SR) can enhance resolution post-scan, yet existing deep learning methods face fidelity-efficiency trade-offs. Purpose: To develop a computationally efficient and accurate deep learning framework for MRI SR that preserves anatomical detail for clinical integration. Materials and Methods: We propose a novel SR framework combining multi-head selective state-space models (MHSSM) with a lightweight channel MLP. The model uses 2D patch extraction with hybrid scanning to capture long-range dependencies. Each MambaFormer block integrates MHSSM, depthwise convolutions, and gated channel mixing. Evaluation used 7T brain T1 MP2RAGE maps (n=142) and 1.5T prostate T2w MRI (n=334). Comparisons included Bicubic interpolation, GANs (CycleGAN, Pix2pix, SPSR), transformers (SwinIR), Mamba (MambaIR), and diffusion models (I2SB, Res-SRDiff). Results: Our model achieved superior performance with exceptional efficiency. For 7T brain data: SSIM=0.951+-0.021, PSNR=26.90+-1.41 dB, LPIPS=0.076+-0.022, GMSD=0.083+-0.017, significantly outperforming all baselines (p<0.001). For prostate data: SSIM=0.770+-0.049, PSNR=27.15+-2.19 dB, LPIPS=0.190+-0.095, GMSD=0.087+-0.013. The framework used only 0.9M parameters and 57 GFLOPs, reducing parameters by 99.8% and computation by 97.5% versus Res-SRDiff, while outperforming SwinIR and MambaIR in accuracy and efficiency. Conclusion: The proposed framework provides an efficient, accurate MRI SR solution, delivering enhanced anatomical detail across datasets. Its low computational demand and state-of-the-art performance show strong potential for clinical translation.
Authors: Hanyang Kong, Xingyi Yang, Xiaoxu Zheng, Xinchao Wang
Abstract: Generating long-range, geometrically consistent video presents a fundamental dilemma: while consistency demands strict adherence to 3D geometry in pixel space, state-of-the-art generative models operate most effectively in a camera-conditioned latent space. This disconnect causes current methods to struggle with occluded areas and complex camera trajectories. To bridge this gap, we propose WorldWarp, a framework that couples a 3D structural anchor with a 2D generative refiner. To establish geometric grounding, WorldWarp maintains an online 3D geometric cache built via Gaussian Splatting (3DGS). By explicitly warping historical content into novel views, this cache acts as a structural scaffold, ensuring each new frame respects prior geometry. However, static warping inevitably leaves holes and artifacts due to occlusions. We address this using a Spatio-Temporal Diffusion (ST-Diff) model designed for a "fill-and-revise" objective. Our key innovation is a spatio-temporal varying noise schedule: blank regions receive full noise to trigger generation, while warped regions receive partial noise to enable refinement. By dynamically updating the 3D cache at every step, WorldWarp maintains consistency across video chunks. Consequently, it achieves state-of-the-art fidelity by ensuring that 3D logic guides structure while diffusion logic perfects texture. Project page: \href{https://hyokong.github.io/worldwarp-page/}{https://hyokong.github.io/worldwarp-page/}.
URLs: https://hyokong.github.io/worldwarp-page/, https://hyokong.github.io/worldwarp-page/
Authors: Xinyao Liao, Qiyuan He, Kai Xu, Xiaoye Qu, Yicong Li, Wei Wei, Angela Yao
Abstract: Autoregressive (AR) visual generation relies on tokenizers to map images to and from discrete sequences. However, tokenizers are trained to reconstruct clean images from ground-truth tokens, while AR generators are optimized only for token likelihood. This misalignment leads to generated token sequences that may decode into low-quality images, without direct supervision from the pixel space. We propose VA-$\pi$, a lightweight post-training framework that directly optimizes AR models with a principled pixel-space objective. VA-$\pi$ formulates the generator-tokenizer alignment as a variational optimization, deriving an evidence lower bound (ELBO) that unifies pixel reconstruction and autoregressive modeling. To optimize under the discrete token space, VA-$\pi$ introduces a reinforcement-based alignment strategy that treats the AR generator as a policy, uses pixel-space reconstruction quality as its intrinsic reward. The reward is measured by how well the predicted token sequences can reconstruct the original image under teacher forcing, giving the model direct pixel-level guidance without expensive free-running sampling. The regularization term of the ELBO serves as a natural regularizer, maintaining distributional consistency of tokens. VA-$\pi$ enables rapid adaptation of existing AR generators, without neither tokenizer retraining nor external reward models. With only 1% ImageNet-1K data and 25 minutes of tuning, it reduces FID from 14.36 to 7.65 and improves IS from 86.55 to 116.70 on LlamaGen-XXL, while also yielding notable gains in the text-to-image task on GenEval for both visual generation model (LlamaGen: from 0.306 to 0.339) and unified multi-modal model (Janus-Pro: from 0.725 to 0.744). Code is available at https://github.com/Lil-Shake/VA-Pi.
Authors: Mingrui Wu, Zhaozhi Wang, Fangjinhua Wang, Jiaolong Yang, Marc Pollefeys, Tong Zhang
Abstract: While Multimodal Large Language Models (MLLMs) have achieved impressive performance on semantic tasks, their spatial intelligence--crucial for robust and grounded AI systems--remains underdeveloped. Existing benchmarks fall short of diagnosing this limitation: they either focus on overly simplified qualitative reasoning or rely on domain-specific indoor data, constrained by the lack of outdoor datasets with verifiable metric ground truth. To bridge this gap, we introduce a large-scale benchmark built from pedestrian-perspective videos captured with synchronized stereo cameras, LiDAR, and IMU/GPS sensors. This dataset provides metrically precise 3D information, enabling the automatic generation of spatial reasoning questions that span a hierarchical spectrum--from qualitative relational reasoning to quantitative metric and kinematic understanding. Evaluations reveal that the performance gains observed in structured indoor benchmarks vanish in open-world settings. Further analysis using synthetic abnormal scenes and blinding tests confirms that current MLLMs depend heavily on linguistic priors instead of grounded visual reasoning. Our benchmark thus provides a principled platform for diagnosing these limitations and advancing physically grounded spatial intelligence.
Authors: Dixuan Lin, Tianyou Wang, Zhuoyang Pan, Yufu Wang, Lingjie Liu, Kostas Daniilidis
Abstract: We build the first system to address the problem of reconstructing in-scene object manipulation from a monocular RGB video. It is challenging due to ill-posed scene reconstruction, ambiguous hand-object depth, and the need for physically plausible interactions. Existing methods operate in hand centric coordinates and ignore the scene, hindering metric accuracy and practical use. In our method, we first use data-driven foundation models to initialize the core components, including the object mesh and poses, the scene point cloud, and the hand poses. We then apply a two-stage optimization that recovers a complete hand-object motion from grasping to interaction, which remains consistent with the scene information observed in the input video.
Authors: Zixuan Ye, Quande Liu, Cong Wei, Yuanxing Zhang, Xintao Wang, Pengfei Wan, Kun Gai, Wenhan Luo
Abstract: Recently, the introduction of Chain-of-Thought (CoT) has largely improved the generation ability of unified models. However, it is observed that the current thinking process during generation mainly focuses on the text consistency with the text prompt, ignoring the \textbf{visual context consistency} with the visual reference images during the multi-modal generation, e.g., multi-reference generation. The lack of such consistency results in the failure in maintaining key visual features (like human ID, object attribute, style). To this end, we integrate the visual context consistency into the reasoning of unified models, explicitly motivating the model to sustain such consistency by 1) Adaptive Visual Planning: generating structured visual check list to figure out the visual element of needed consistency keeping, and 2) Iterative Visual Correction: performing self-reflection with the guidance of check lists and refining the generated result in an iterative manner. To achieve this, we use supervised finetuning to teach the model how to plan the visual checking, conduct self-reflection and self-refinement, and use flow-GRPO to further enhance the visual consistency through a customized visual checking reward. The experiments show that our method outperforms both zero-shot unified models and those with text CoTs in multi-modal generation, demonstrating higher visual context consistency.
Authors: Pablo Ruiz-Ponce, Sergio Escalera, Jos\'e Garc\'ia-Rodr\'iguez, Jiankang Deng, Rolandos Alexandros Potamias
Abstract: Generating realistic human-human interactions is a challenging task that requires not only high-quality individual body and hand motions, but also coherent coordination among all interactants. Due to limitations in available data and increased learning complexity, previous methods tend to ignore hand motions, limiting the realism and expressivity of the interactions. Additionally, current diffusion-based approaches generate entire motion sequences simultaneously, limiting their ability to capture the reactive and adaptive nature of human interactions. To address these limitations, we introduce Interact2Ar, the first end-to-end text-conditioned autoregressive diffusion model for generating full-body, human-human interactions. Interact2Ar incorporates detailed hand kinematics through dedicated parallel branches, enabling high-fidelity full-body generation. Furthermore, we introduce an autoregressive pipeline coupled with a novel memory technique that facilitates adaptation to the inherent variability of human interactions using efficient large context windows. The adaptability of our model enables a series of downstream applications, including temporal motion composition, real-time adaptation to disturbances, and extension beyond dyadic to multi-person scenarios. To validate the generated motions, we introduce a set of robust evaluators and extended metrics designed specifically for assessing full-body interactions. Through quantitative and qualitative experiments, we demonstrate the state-of-the-art performance of Interact2Ar.
Authors: Weichen Fan, Haiwen Diao, Quan Wang, Dahua Lin, Ziwei Liu
Abstract: Deep representations across modalities are inherently intertwined. In this paper, we systematically analyze the spectral characteristics of various semantic and pixel encoders. Interestingly, our study uncovers a highly inspiring and rarely explored correspondence between an encoder's feature spectrum and its functional role: semantic encoders primarily capture low-frequency components that encode abstract meaning, whereas pixel encoders additionally retain high-frequency information that conveys fine-grained detail. This heuristic finding offers a unifying perspective that ties encoder behavior to its underlying spectral structure. We define it as the Prism Hypothesis, where each data modality can be viewed as a projection of the natural world onto a shared feature spectrum, just like the prism. Building on this insight, we propose Unified Autoencoding (UAE), a model that harmonizes semantic structure and pixel details via an innovative frequency-band modulator, enabling their seamless coexistence. Extensive experiments on ImageNet and MS-COCO benchmarks validate that our UAE effectively unifies semantic abstraction and pixel-level fidelity into a single latent space with state-of-the-art performance.
Authors: Bangya Liu, Chengpo Yan, Chenghao Jiang, Suman Banerjee, Akarsh Prabhakara
Abstract: Cooperative perception between vehicles is poised to offer robust and reliable scene understanding. Recently, we are witnessing experimental systems research building testbeds that share raw spatial sensor data for cooperative perception. While there has been a marked improvement in accuracies and is the natural way forward, we take a moment to consider the problems with such an approach for eventual adoption by automakers. In this paper, we first argue that new forms of privacy concerns arise and discourage stakeholders to share raw sensor data. Next, we present SHARP, a research framework to minimize privacy leakage and drive stakeholders towards the ambitious goal of raw data based cooperative perception. Finally, we discuss open questions for networked systems, mobile computing, perception researchers, industry and government in realizing our proposed framework.
Authors: Benjamin S. H. Lyo, Eero P. Simoncelli, Cristina Savin
Abstract: We present the Score-based Autoencoder for Multiscale Inference (SAMI), a method for unsupervised representation learning that combines the theoretical frameworks of diffusion models and VAEs. By unifying their respective evidence lower bounds, SAMI formulates a principled objective that learns representations through score-based guidance of the underlying diffusion process. The resulting representations automatically capture meaningful structure in the data: it recovers ground truth generative factors in our synthetic dataset, learns factorized, semantic latent dimensions from complex natural images, and encodes video sequences into latent trajectories that are straighter than those of alternative encoders, despite training exclusively on static images. Furthermore, SAMI can extract useful representations from pre-trained diffusion models with minimal additional training. Finally, the explicitly probabilistic formulation provides new ways to identify semantically meaningful axes in the absence of supervised labels, and its mathematical exactness allows us to make formal statements about the nature of the learned representation. Overall, these results indicate that implicit structural information in diffusion models can be made explicit and interpretable through synergistic combination with a variational autoencoder.
Authors: Viv Atureta, Rifki Priansyah Jasin, Stefan Siegert
Abstract: This paper documents a data set of UK rain radar image sequences for use in statistical modeling and machine learning methods for nowcasting. The main dataset contains 1,000 randomly sampled sequences of length 20 steps (15-minute increments) of 2D radar intensity fields of dimension 40x40 (at 5km spatial resolution). Spatially stratified sampling ensures spatial homogeneity despite removal of clear-sky cases by threshold-based truncation. For each radar sequence, additional atmospheric and geographic features are made available, including date, location, mean elevation, mean wind direction and speed and prevailing storm type. New R functions to extract data from the binary "Nimrod" radar data format are provided. A case study is presented to train and evaluate a simple convolutional neural network for radar nowcasting, including self-contained R code.
Authors: Aziz Muminov, Anne Pham
Abstract: Bone marrow cell cytomorphology analysis is critical for the diagnosis of hematological malignancies but remains a labor-intensive process subject to significant inter-observer variability. While recent foundation models have shown promise in computational pathology, they often require extensive computational resources and fail to account for the asymmetric risks associated with clinical misdiagnosis. We introduce CytoDINO, a framework that achieves state-of-the-art performance on the Munich Leukemia Laboratory (MLL) dataset by fine-tuning DINOv3 using Low-Rank Adaptation (LoRA). Our primary contribution is a novel Hierarchical Focal Loss with Critical Penalties, which encodes biological relationships between cell lineages and explicitly penalizes clinically dangerous misclassifications (e.g., classifying blasts as normal cells). CytoDINO achieves an 88.2% weighted F1 score and 76.5% macro F1 on a held-out test set of 21 cell classes. By utilizing parameter-efficient fine-tuning with only 8% trainable parameters on a single NVIDIA RTX 5080, we demonstrate that consumer-grade hardware can match specialized infrastructure. Furthermore, confidence-based selective prediction yields 99.5% accuracy on 67% of samples, suggesting a viable pathway for clinical deployment where high-uncertainty cases are flagged for expert review
Authors: Yu Fang, Kanchana Ranasinghe, Le Xue, Honglu Zhou, Juntao Tan, Ran Xu, Shelby Heinecke, Caiming Xiong, Silvio Savarese, Daniel Szafir, Mingyu Ding, Michael S. Ryoo, Juan Carlos Niebles
Abstract: Vision-Language-Action (VLA) models have achieved remarkable progress in robotic manipulation by mapping multimodal observations and instructions directly to actions. However, they typically mimic expert trajectories without predictive motion reasoning, which limits their ability to reason about what actions to take. To address this limitation, we propose joint learning with motion image diffusion, a novel strategy that enhances VLA models with motion reasoning capabilities. Our method extends the VLA architecture with a dual-head design: while the action head predicts action chunks as in vanilla VLAs, an additional motion head, implemented as a Diffusion Transformer (DiT), predicts optical-flow-based motion images that capture future dynamics. The two heads are trained jointly, enabling the shared VLM backbone to learn representations that couple robot control with motion knowledge. This joint learning builds temporally coherent and physically grounded representations without modifying the inference pathway of standard VLAs, thereby maintaining test-time latency. Experiments in both simulation and real-world environments demonstrate that joint learning with motion image diffusion improves the success rate of pi-series VLAs to 97.5% on the LIBERO benchmark and 58.0% on the RoboTwin benchmark, yielding a 23% improvement in real-world performance and validating its effectiveness in enhancing the motion reasoning capability of large-scale VLAs.
Authors: Tin Stribor Sohn, Maximilian Dillitzer, Jason J. Corso, Eric Sax
Abstract: Vision-language navigation requires agents to reason and act under constraints of embodiment. While vision-language models (VLMs) demonstrate strong generalization, current benchmarks provide limited understanding of how embodiment -- i.e., the choice of physical platform, sensor configuration, and modality alignment -- influences perception, reasoning, and control. We introduce Embodied4C, a closed-loop benchmark designed as a Turing test for embodied reasoning. The benchmark evaluates the core embodied capabilities of VLMs across three heterogeneous embodiments -- autonomous vehicles, aerial drones, and robotic manipulators -- through approximately 1.1K one-shot reasoning questions and 58 goal-directed navigation tasks. These tasks jointly assess four foundational dimensions: semantic, spatial, temporal, and physical reasoning. Each embodiment presents dynamic sensor configurations and environment variations to probe generalization beyond platform-specific adaptation. To prevent embodiment overfitting, Embodied4C integrates domain-far queries targeting abstract and cross-context reasoning. Comprehensive evaluation across ten state-of-the-art VLMs and four embodied control baselines shows that cross-modal alignment and instruction tuning matter more than scale, while spatial and temporal reasoning remains the primary bottleneck for reliable embodied competence.
Authors: Changxu Duan
Abstract: Academic documents stored in PDF format can be transformed into plain text structured markup languages to enhance accessibility and enable scalable digital library workflows. Markup languages allow for easier updates and customization, making academic content more adaptable and accessible to diverse usage, such as linguistic corpus compilation. Such documents, typically delivered in PDF format, contain complex elements including mathematical formulas, figures, headers, and tables, as well as densely layouted text. Existing end-to-end decoder transformer models can transform screenshots of documents into markup language. However, these models exhibit significant inefficiencies; their token-by-token decoding from scratch wastes a lot of inference steps in regenerating dense text that could be directly copied from PDF files. To solve this problem, we introduce EditTrans, a hybrid editing-generation model whose features allow identifying a queue of to-be-edited text from a PDF before starting to generate markup language. EditTrans contains a lightweight classifier fine-tuned from a Document Layout Analysis model on 162,127 pages of documents from arXiv. In our evaluations, EditTrans reduced the transformation latency up to 44.5% compared to end-to-end decoder transformer models, while maintaining transformation quality. Our code and reproducible dataset production scripts are open-sourced.
Authors: Midhat Urooj, Ayan Banerjee, Sandeep Gupta
Abstract: Accurate yet interpretable image-based diagnosis remains a central challenge in medical AI, particularly in settings characterized by limited data, subtle visual cues, and high-stakes clinical decision-making. Most existing vision models rely on purely data-driven learning and produce black-box predictions with limited interpretability and poor cross-domain generalization, hindering their real-world clinical adoption. We present NEURO-GUARD, a novel knowledge-guided vision framework that integrates Vision Transformers (ViTs) with language-driven reasoning to improve performance, transparency, and domain robustness. NEURO-GUARD employs a retrieval-augmented generation (RAG) mechanism for self-verification, in which a large language model (LLM) iteratively generates, evaluates, and refines feature-extraction code for medical images. By grounding this process in clinical guidelines and expert knowledge, the framework progressively enhances feature detection and classification beyond purely data-driven baselines. Extensive experiments on diabetic retinopathy classification across four benchmark datasets APTOS, EyePACS, Messidor-1, and Messidor-2 demonstrate that NEURO-GUARD improves accuracy by 6.2% over a ViT-only baseline (84.69% vs. 78.4%) and achieves a 5% gain in domain generalization. Additional evaluations on MRI-based seizure detection further confirm its cross-domain robustness, consistently outperforming existing methods. Overall, NEURO-GUARD bridges symbolic medical reasoning with subsymbolic visual learning, enabling interpretable, knowledge-aware, and generalizable medical image diagnosis while achieving state-of-the-art performance across multiple datasets.
Authors: Yilei Wu, Yichi Zhang, Zijian Dong, Fang Ji, An Sen Tan, Gifford Tan, Sizhao Tang, Huijuan Chen, Zijiao Chen, Eric Kwun Kei Ng, Jose Bernal, Hang Min, Ying Xia, Ines Vati, Liz Cooper, Xiaoyu Hu, Yuchen Pei, Yutao Ma, Victor Nozais, Ami Tsuchida, Pierre-Yves Herv\'e, Philippe Boutinaud, Marc Joliot, Junghwa Kang, Wooseung Kim, Dayeon Bak, Rachika E. Hamadache, Valeriia Abramova, Xavier Llad\'o, Yuntao Zhu, Zhenyu Gong, Xin Chen, John McFadden, Pek Lan Khong, Roberto Duarte Coello, Hongwei Bran Li, Woon Puay Koh, Christopher Chen, Joanna M. Wardlaw, Maria del C. Vald\'es Hern\'andez, Juan Helen Zhou
Abstract: Perivascular spaces (PVS), when abnormally enlarged and visible in magnetic resonance imaging (MRI) structural sequences, are important imaging markers of cerebral small vessel disease and potential indicators of neurodegenerative conditions. Despite their clinical significance, automatic enlarged PVS (EPVS) segmentation remains challenging due to their small size, variable morphology, similarity with other pathological features, and limited annotated datasets. This paper presents the EPVS Challenge organized at MICCAI 2024, which aims to advance the development of automated algorithms for EPVS segmentation across multi-site data. We provided a diverse dataset comprising 100 training, 50 validation, and 50 testing scans collected from multiple international sites (UK, Singapore, and China) with varying MRI protocols and demographics. All annotations followed the STRIVE protocol to ensure standardized ground truth and covered the full brain parenchyma. Seven teams completed the full challenge, implementing various deep learning approaches primarily based on U-Net architectures with innovations in multi-modal processing, ensemble strategies, and transformer-based components. Performance was evaluated using dice similarity coefficient, absolute volume difference, recall, and precision metrics. The winning method employed MedNeXt architecture with a dual 2D/3D strategy for handling varying slice thicknesses. The top solutions showed relatively good performance on test data from seen datasets, but significant degradation of performance was observed on the previously unseen Shanghai cohort, highlighting cross-site generalization challenges due to domain shift. This challenge establishes an important benchmark for EPVS segmentation methods and underscores the need for the continued development of robust algorithms that can generalize in diverse clinical settings.
Authors: Hyeonjin Lee, Jun-Hyuk Kim, Jong-Seok Lee
Abstract: In recent years, the demand of image compression models for machine vision has increased dramatically. However, the training frameworks of image compression still focus on the vision of human, maintaining the excessive perceptual details, thus have limitations in optimally reducing the bits per pixel in the case of performing machine vision tasks. In this paper, we propose Semantic-based Low-bitrate Image compression for Machines by leveraging diffusion, termed SLIM. This is a new effective training framework of image compression for machine vision, using a pretrained latent diffusion model.The compressor model of our method focuses only on the Region-of-Interest (RoI) areas for machine vision in the image latent, to compress it compactly. Then the pretrained Unet model enhances the decompressed latent, utilizing a RoI-focused text caption which containing semantic information of the image. Therefore, SLIM is able to focus on RoI areas of the image without any guide mask at the inference stage, achieving low bitrate when compressing. And SLIM is also able to enhance a decompressed latent by denoising steps, so the final reconstructed image from the enhanced latent can be optimized for the machine vision task while still containing perceptual details for human vision. Experimental results show that SLIM achieves a higher classification accuracy in the same bits per pixel condition, compared to conventional image compression models for machines.Code will be released upon acceptance.
Authors: Rui Liu, Dian Yu, Lei Ke, Haolin Liu, Yujun Zhou, Zhenwen Liang, Haitao Mi, Pratap Tokekar, Dong Yu
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has become a key paradigm to improve the reasoning capabilities of Multimodal Large Language Models (MLLMs). However, prevalent group-based algorithms such as GRPO require multi-rollout sampling for each prompt. While more efficient single-rollout variants have recently been explored in text-only settings, we find that they suffer from severe instability in multimodal contexts, often leading to training collapse. To address this training efficiency-stability trade-off, we introduce $\textbf{MSSR}$ (Multimodal Stabilized Single-Rollout), a group-free RLVR framework that achieves both stable optimization and effective multimodal reasoning performance. MSSR achieves this via an entropy-based advantage-shaping mechanism that adaptively regularizes advantage magnitudes, preventing collapse and maintaining training stability. While such mechanisms have been used in group-based RLVR, we show that in the multimodal single-rollout setting they are not merely beneficial but essential for stability. In in-distribution evaluations, MSSR demonstrates superior training compute efficiency, achieving similar validation accuracy to the group-based baseline with half the training steps. When trained for the same number of steps, MSSR's performance surpasses the group-based baseline and shows consistent generalization improvements across five diverse reasoning-intensive benchmarks. Together, these results demonstrate that MSSR enables stable, compute-efficient, and effective RLVR for complex multimodal reasoning tasks.
Authors: Eren Caglar, Amirkia Rafiei Oskooei, Mehmet Kutanoglu, Mustafa Keles, Mehmet S. Aktas
Abstract: This paper introduces a parallel and asynchronous Transformer framework designed for efficient and accurate multilingual lip synchronization in real-time video conferencing systems. The proposed architecture integrates translation, speech processing, and lip-synchronization modules within a pipeline-parallel design that enables concurrent module execution through message-queue-based decoupling, reducing end-to-end latency by up to 3.1 times compared to sequential approaches. To enhance computational efficiency and throughput, the inference workflow of each module is optimized through low-level graph compilation, mixed-precision quantization, and hardware-accelerated kernel fusion. These optimizations provide substantial gains in efficiency while preserving model accuracy and visual quality. In addition, a context-adaptive silence-detection component segments the input speech stream at semantically coherent boundaries, improving translation consistency and temporal alignment across languages. Experimental results demonstrate that the proposed parallel architecture outperforms conventional sequential pipelines in processing speed, synchronization stability, and resource utilization. The modular, message-oriented design makes this work applicable to resource-constrained IoT communication scenarios including telemedicine, multilingual kiosks, and remote assistance systems. Overall, this work advances the development of low-latency, resource-efficient multimodal communication frameworks for next-generation AIoT systems.
Authors: Xavier Rafael-Palou, Jose Munuera, Ana Jimenez-Pastor, Richard Osuala, Karim Lekadir, Oliver Diaz
Abstract: Modern clinical decision support systems can concurrently serve multiple, independent medical imaging institutions, but their predictive performance may degrade across sites due to variations in patient populations, imaging hardware, and acquisition protocols. Continuous surveillance of predictive model outputs offers a safe and reliable approach for identifying such distributional shifts without ground truth labels. However, most existing methods rely on centralized monitoring of aggregated predictions, overlooking site-specific drift dynamics. We propose an agent-based framework for detecting drift and assessing its severity in multisite clinical AI systems. To evaluate its effectiveness, we simulate a multi-center environment for output-based drift detection, assigning each site a drift monitoring agent that performs batch-wise comparisons of model outputs against a reference distribution. We analyse several multi-center monitoring schemes, that differ in how the reference is obtained (site-specific, global, production-only and adaptive), alongside a centralized baseline. Results on real-world breast cancer imaging data using a pathological complete response prediction model shows that all multi-center schemes outperform centralized monitoring, with F1-score improvements up to 10.3% in drift detection. In the absence of site-specific references, the adaptive scheme performs best, with F1-scores of 74.3% for drift detection and 83.7% for drift severity classification. These findings suggest that adaptive, site-aware agent-based drift monitoring can enhance reliability of multisite clinical decision support systems.
Authors: Jayant Lohia
Abstract: Winograd convolution is the standard algorithm for efficient inference, reducing arithmetic complexity by 2.25x for 3x3 kernels. However, it faces a critical barrier in the modern era of low precision computing: numerical instability. As tiles scale to maximize efficiency (e.g., F(6,3), F(8,3)), the condition numbers of standard integer based transforms explode, reaching kappa = 2 x 10^5 for F(8,3), rendering them unusable in FP16 or Int8. We introduce NOVA (Numerical Optimization of Vandermonde Arithmetic), a discovery framework that breaks the decades old convention of integer interpolation. Treating Winograd point selection as a continuous optimization problem, NOVA searches the manifold R^n-1 via Evolution Strategy, snaps candidates to simple rationals, and guarantees correctness via symbolic verification. This process uncovers a hidden landscape of stable, fractional configurations such as {+-5/6, +-7/6, +-3/5} that defy traditional vocabulary constraints. The impact is transformative: NOVA improves the conditioning of F(8,3) by 415x in 1D, which squares to a 172,484x improvement for 2D convolution. In real world FP16 ImageNet inference, where standard transforms collapse to random chance (e.g., 4.7 percent accuracy on VGG16), NOVA's points restore full accuracy (75 to 78 percent), recovering over 70 percentage points without retraining, calibration, or learned parameters. These discovered transforms act as drop in replacements, effectively unlocking the efficiency of large tile Winograd convolution for next generation hardware.
Authors: Wenjun Lin, Jensen Zhang, Kaitong Cai, Keze Wang
Abstract: We present STORM (Search-Guided Generative World Models), a novel framework for spatio-temporal reasoning in robotic manipulation that unifies diffusion-based action generation, conditional video prediction, and search-based planning. Unlike prior Vision-Language-Action (VLA) models that rely on abstract latent dynamics or delegate reasoning to language components, STORM grounds planning in explicit visual rollouts, enabling interpretable and foresight-driven decision-making. A diffusion-based VLA policy proposes diverse candidate actions, a generative video world model simulates their visual and reward outcomes, and Monte Carlo Tree Search (MCTS) selectively refines plans through lookahead evaluation. Experiments on the SimplerEnv manipulation benchmark demonstrate that STORM achieves a new state-of-the-art average success rate of 51.0 percent, outperforming strong baselines such as CogACT. Reward-augmented video prediction substantially improves spatio-temporal fidelity and task relevance, reducing Frechet Video Distance by over 75 percent. Moreover, STORM exhibits robust re-planning and failure recovery behavior, highlighting the advantages of search-guided generative world models for long-horizon robotic manipulation.
Authors: Weijie Zhou, Xuangtang Xiong, Ye Tian, Lijun Yue, Xinyu Wu, Wei Li, Chaoyang Zhao, Honghui Dong, Ming Tang, Jinqiao Wang, Zhengyou Zhang
Abstract: Multimodal Large Language Models (MLLMs) have empowered embodied agents with remarkable capabilities in planning and reasoning. However, when facing ambiguous natural language instructions (e.g., "fetch the tool" in a cluttered room), current agents often fail to balance the high cost of physical exploration against the cognitive cost of human interaction. They typically treat disambiguation as a passive perception problem, lacking the strategic reasoning to minimize total task execution costs. To bridge this gap, we propose ESearch-R1, a cost-aware embodied reasoning framework that unifies interactive dialogue (Ask), episodic memory retrieval (GetMemory), and physical navigation (Navigate) into a single decision process. We introduce HC-GRPO (Heterogeneous Cost-Aware Group Relative Policy Optimization). Unlike traditional PPO which relies on a separate value critic, HC-GRPO optimizes the MLLM by sampling groups of reasoning trajectories and reinforcing those that achieve the optimal trade-off between information gain and heterogeneous costs (e.g., navigate time, and human attention). Extensive experiments in AI2-THOR demonstrate that ESearch-R1 significantly outperforms standard ReAct-based agents. It improves task success rates while reducing total operational costs by approximately 50\%, validating the effectiveness of GRPO in aligning MLLM agents with physical world constraints.
Authors: Chihiro Noguchi, Takaki Yamamoto
Abstract: End-to-end (E2E) autonomous driving models that take only camera images as input and directly predict a future trajectory are appealing for their computational efficiency and potential for improved generalization via unified optimization; however, persistent failure modes remain due to reliance on imitation learning (IL). While online reinforcement learning (RL) could mitigate IL-induced issues, the computational burden of neural rendering-based simulation and large E2E networks renders iterative reward and hyperparameter tuning costly. We introduce a camera-only E2E offline RL framework that performs no additional exploration and trains solely on a fixed simulator dataset. Offline RL offers strong data efficiency and rapid experimental iteration, yet is susceptible to instability from overestimation on out-of-distribution (OOD) actions. To address this, we construct pseudo ground-truth trajectories from expert driving logs and use them as a behavior regularization signal, suppressing imitation of unsafe or suboptimal behavior while stabilizing value learning. Training and closed-loop evaluation are conducted in a neural rendering environment learned from the public nuScenes dataset. Empirically, the proposed method achieves substantial improvements in collision rate and route completion compared with IL baselines. Our code will be available at [URL].
Authors: Ryosuke Korekata, Quanting Xie, Yonatan Bisk, Komei Sugiura
Abstract: In this study, we address the problem of open-vocabulary mobile manipulation, where a robot is required to carry a wide range of objects to receptacles based on free-form natural language instructions. This task is challenging, as it involves understanding visual semantics and the affordance of manipulation actions. To tackle these challenges, we propose Affordance RAG, a zero-shot hierarchical multimodal retrieval framework that constructs Affordance-Aware Embodied Memory from pre-explored images. The model retrieves candidate targets based on regional and visual semantics and reranks them with affordance scores, allowing the robot to identify manipulation options that are likely to be executable in real-world environments. Our method outperformed existing approaches in retrieval performance for mobile manipulation instruction in large-scale indoor environments. Furthermore, in real-world experiments where the robot performed mobile manipulation in indoor environments based on free-form instructions, the proposed method achieved a task success rate of 85%, outperforming existing methods in both retrieval performance and overall task success.
Authors: Pengxuan Yang, Ben Lu, Zhongpu Xia, Chao Han, Yinfeng Gao, Teng Zhang, Kun Zhan, XianPeng Lang, Yupeng Zheng, Qichao Zhang
Abstract: Latent World Models enhance scene representation through temporal self-supervised learning, presenting a perception annotation-free paradigm for end-to-end autonomous driving. However, the reconstruction-oriented representation learning tangles perception with planning tasks, leading to suboptimal optimization for planning. To address this challenge, we propose WorldRFT, a planning-oriented latent world model framework that aligns scene representation learning with planning via a hierarchical planning decomposition and local-aware interactive refinement mechanism, augmented by reinforcement learning fine-tuning (RFT) to enhance safety-critical policy performance. Specifically, WorldRFT integrates a vision-geometry foundation model to improve 3D spatial awareness, employs hierarchical planning task decomposition to guide representation optimization, and utilizes local-aware iterative refinement to derive a planning-oriented driving policy. Furthermore, we introduce Group Relative Policy Optimization (GRPO), which applies trajectory Gaussianization and collision-aware rewards to fine-tune the driving policy, yielding systematic improvements in safety. WorldRFT achieves state-of-the-art (SOTA) performance on both open-loop nuScenes and closed-loop NavSim benchmarks. On nuScenes, it reduces collision rates by 83% (0.30% -> 0.05%). On NavSim, using camera-only sensors input, it attains competitive performance with the LiDAR-based SOTA method DiffusionDrive (87.8 vs. 88.1 PDMS).
Authors: Dazhen Deng, Sen Yang, Yuchen He, Yuan Tian, Yingcai Wu
Abstract: Current chart-specific tasks, such as chart question answering, chart parsing, and chart generation, are typically studied in isolation, preventing models from learning the shared semantics that link chart generation and interpretation. We introduce CycleChart, a consistency-based learning framework for bidirectional chart understanding and generation. CycleChart adopts a schema-centric formulation as a common interface across tasks. We construct a consistent multi-task dataset, where each chart sample includes aligned annotations for schema prediction, data parsing, and question answering. To learn cross-directional chart semantics, CycleChart introduces a generate-parse consistency objective: the model generates a chart schema from a table and a textual query, then learns to recover the schema and data from the generated chart, enforcing semantic alignment across directions. CycleChart achieves strong results on chart generation, chart parsing, and chart question answering, demonstrating improved cross-task generalization and marking a step toward more general chart understanding models.
Authors: Beyza Zayim, Aissiou Ikram, Boukhiar Naima
Abstract: Breast cancer remains the most common cancer among women and is a leading cause of female mortality. Dynamic contrast-enhanced MRI (DCE-MRI) is a powerful imaging tool for evaluating breast tumors, yet the field lacks a standardized benchmark for analyzing treatment responses and guiding personalized care. We participated in the MAMA-MIA Challenge's Primary Tumor Segmentation task and this work presents a proposed selective, phase-aware training framework for the nnU-Net architecture, emphasizing quality-focused data selection to strengthen model robustness and generalization. We employed the No New Net (nnU-Net) framework with a selective training strategy that systematically analyzed the impact of image quality and center-specific variability on segmentation performance. Controlled experiments on the DUKE, NACT, ISPY1, and ISPY2 datasets revealed that including ISPY scans with motion artifacts and reduced contrast impaired segmentation performance, even with advanced preprocessing, such as contrast-limited adaptive histogram equalization (CLAHE). In contrast, training on DUKE and NACT data, which exhibited clearer contrast and fewer motion artifacts despite varying resolutions, with early phase images (0000-0002) provided more stable training conditions. Our results demonstrate the importance of phase-sensitive and quality-aware training strategies in achieving reliable segmentation performance in heterogeneous clinical datasets, highlighting the limitations of the expansion of naive datasets and motivating the need for future automation of quality-based data selection strategies.
Authors: Carla Crivoi, Radu Tudor Ionescu
Abstract: We present the first comprehensive empirical study of machine unlearning (MU) in hybrid quantum-classical neural networks. While MU has been extensively explored in classical deep learning, its behavior within variational quantum circuits (VQCs) and quantum-augmented architectures remains largely unexplored. First, we adapt a broad suite of unlearning methods to quantum settings, including gradient-based, distillation-based, regularization-based and certified techniques. Second, we introduce two new unlearning strategies tailored to hybrid models. Experiments across Iris, MNIST, and Fashion-MNIST, under both subset removal and full-class deletion, reveal that quantum models can support effective unlearning, but outcomes depend strongly on circuit depth, entanglement structure, and task complexity. Shallow VQCs display high intrinsic stability with minimal memorization, whereas deeper hybrid models exhibit stronger trade-offs between utility, forgetting strength, and alignment with retrain oracle. We find that certain methods, e.g. EU-k, LCA, and Certified Unlearning, consistently provide the best balance across metrics. These findings establish baseline empirical insights into quantum machine unlearning and highlight the need for quantum-aware algorithms and theoretical guarantees, as quantum machine learning systems continue to expand in scale and capability. We publicly release our code at: https://github.com/CrivoiCarla/HQML.
Authors: Yayuan Li, Jian Zhang, Jintao Guo, Zihan Cheng, Lei Qi, Yinghuan Shi, Yang Gao
Abstract: The proliferation of pre-trained models has given rise to a wide array of specialised, fine-tuned models. Model merging aims to merge the distinct capabilities of these specialised models into a unified model, requiring minimal or even no additional training. A core objective of model merging is to ensure the merged model retains the behavioural characteristics of the specialised models, typically achieved through feature alignment. We identify that features consist of two critical components: direction and magnitude. Prior research has predominantly focused on directional alignment, while the influence of magnitude remains largely neglected, despite its pronounced vulnerability to perturbations introduced by common merging operations (e.g., parameter fusion and sparsification). Such perturbations to magnitude inevitably lead to feature deviations in the merged model from the specialised models, resulting in subsequent performance degradation. To address this, we propose MAGnItude Calibration (MAGIC), a plug-and-play framework that rectifies layer-wise magnitudes in feature and weight spaces, with three variants. Specifically, our Feature Space Calibration (FSC) realigns the merged model's features using a small set of unlabelled data, while Weight Space Calibration (WSC) extends this calibration to the weight space without requiring additional data. Combining these yields Dual Space Calibration (DSC). Comprehensive experiments demonstrate that MAGIC consistently boosts performance across diverse Computer Vision tasks (+4.3% on eight datasets) and NLP tasks (+8.0% on Llama) without additional training. Our code is available at: https://github.com/lyymuwu/MAGIC
Authors: Hongwei Fan, Hang Dai, Jiyao Zhang, Jinzhou Li, Qiyang Yan, Yujie Zhao, Mingju Gao, Jinghang Wu, Hao Tang, Hao Dong
Abstract: The robotics field is evolving towards data-driven, end-to-end learning, inspired by multimodal large models. However, reliance on expensive real-world data limits progress. Simulators offer cost-effective alternatives, but the gap between simulation and reality challenges effective policy transfer. This paper introduces TwinAligner, a novel Real2Sim2Real system that addresses both visual and dynamic gaps. The visual alignment module achieves pixel-level alignment through SDF reconstruction and editable 3DGS rendering, while the dynamic alignment module ensures dynamic consistency by identifying rigid physics from robot-object interaction. TwinAligner improves robot learning by providing scalable data collection and establishing a trustworthy iterative cycle, accelerating algorithm development. Quantitative evaluations highlight TwinAligner's strong capabilities in visual and dynamic real-to-sim alignment. This system enables policies trained in simulation to achieve strong zero-shot generalization to the real world. The high consistency between real-world and simulated policy performance underscores TwinAligner's potential to advance scalable robot learning. Code and data will be released on https://twin-aligner.github.io
Authors: Yujie Zhao, Hongwei Fan, Di Chen, Shengcong Chen, Liliang Chen, Xiaoqi Li, Guanghui Ren, Hao Dong
Abstract: Recent progress in robot learning has been driven by large-scale datasets and powerful visuomotor policy architectures, yet policy robustness remains limited by the substantial cost of collecting diverse demonstrations, particularly for spatial generalization in manipulation tasks. To reduce repetitive data collection, we present Real2Edit2Real, a framework that generates new demonstrations by bridging 3D editability with 2D visual data through a 3D control interface. Our approach first reconstructs scene geometry from multi-view RGB observations with a metric-scale 3D reconstruction model. Based on the reconstructed geometry, we perform depth-reliable 3D editing on point clouds to generate new manipulation trajectories while geometrically correcting the robot poses to recover physically consistent depth, which serves as a reliable condition for synthesizing new demonstrations. Finally, we propose a multi-conditional video generation model guided by depth as the primary control signal, together with action, edge, and ray maps, to synthesize spatially augmented multi-view manipulation videos. Experiments on four real-world manipulation tasks demonstrate that policies trained on data generated from only 1-5 source demonstrations can match or outperform those trained on 50 real-world demonstrations, improving data efficiency by up to 10-50x. Moreover, experimental results on height and texture editing demonstrate the framework's flexibility and extensibility, indicating its potential to serve as a unified data generation framework.
Authors: Meng Ding, Xiao Fu
Abstract: This work revisits the hyperspectral super-resolution (HSR) problem, i.e., fusing a pair of spatially co-registered hyperspectral (HSI) and multispectral (MSI) images to recover a super-resolution image (SRI) that enhances the spatial resolution of the HSI. Coupled tensor decomposition (CTD)-based methods have gained traction in this domain, offering recoverability guarantees under various assumptions. Existing models such as canonical polyadic decomposition (CPD) and Tucker decomposition provide strong expressive power but lack physical interpretability. The block-term decomposition model with rank-$(L_r, L_r, 1)$ terms (the LL1 model) yields interpretable factors under the linear mixture model (LMM) of spectral images, but LMM assumptions are often violated in practice -- primarily due to nonlinear effects such as endmember variability (EV). To address this, we propose modeling spectral images using a more flexible block-term tensor decomposition with rank-$(L_r, M_r, N_r)$ terms (the LMN model). This modeling choice retains interpretability, subsumes CPD, Tucker, and LL1 as special cases, and robustly accounts for non-ideal effects such as EV, offering a balanced tradeoff between expressiveness and interpretability for HSR. Importantly, under the LMN model for HSI and MSI, recoverability of the SRI can still be established under proper conditions -- providing strong theoretical support. Extensive experiments on synthetic and real datasets further validate the effectiveness and robustness of the proposed method compared with existing CTD-based approaches.
Authors: Eric Guzman, Joel Meyers
Abstract: The search for primordial gravitational waves is a central goal of cosmic microwave background (CMB) surveys. Isolating the characteristic $B$-mode polarization signal sourced by primordial gravitational waves is challenging for several reasons: the amplitude of the signal is inherently small; astrophysical foregrounds produce $B$-mode polarization contaminating the signal; and secondary $B$-mode polarization fluctuations are produced via the conversion of $E$ modes. Current and future low-noise, multi-frequency observations enable sufficient precision to address the first two of these challenges such that secondary $B$ modes will become the bottleneck for improved constraints on the amplitude of primordial gravitational waves. The dominant source of secondary $B$-mode polarization is gravitational lensing by large scale structure. Various strategies have been developed to estimate the lensing deflection and to reverse its effects the CMB, thus reducing confusion from lensing $B$ modes in the search for primordial gravitational waves. However, a few complications remain. First, there may be additional sources of secondary $B$-mode polarization, for example from patchy reionization or from cosmic polarization rotation. Second, the statistics of delensed CMB maps can become complicated and non-Gaussian, especially when advanced lensing reconstruction techniques are applied. We previously demonstrated how a deep learning network, ResUNet-CMB, can provide nearly optimal simultaneous estimates of multiple sources of secondary $B$-mode polarization. In this paper, we show how deep learning can be applied to estimate and remove multiple sources of secondary $B$-mode polarization, and we further show how this technique can be used in a likelihood analysis to produce nearly optimal, unbiased estimates of the amplitude of primordial gravitational waves.
Authors: Ziqian Huang, Boxiao Yu, Siqi Li, Savas Ozdemir, Sangjin Bae, Jae Sung Lee, Guobao Wang, Kuang Gong
Abstract: Dynamic PET enables the quantitative estimation of physiology-related parameters and is widely utilized in research and increasingly adopted in clinical settings. Parametric imaging in dynamic PET requires kinetic modeling to estimate voxel-wise physiological parameters based on specific kinetic models. However, parametric images estimated through kinetic model fitting often suffer from low image quality due to the inherently ill-posed nature of the fitting process and the limited counts resulting from non-continuous data acquisition across multiple bed positions in whole-body PET. In this work, we proposed a diffusion model-based kinetic modeling framework for parametric image estimation, using the Patlak model as an example. The score function of the diffusion model was pre-trained on static total-body PET images and served as a prior for both Patlak slope and intercept images by leveraging their patch-wise similarity. During inference, the kinetic model was incorporated as a data-consistency constraint to guide the parametric image estimation. The proposed framework was evaluated on total-body dynamic PET datasets with different dose levels, demonstrating the feasibility and promising performance of the proposed framework in improving parametric image quality.
Authors: Eric Zimmermann, Harley Wiltzer, Justin Szeto, David Alvarez-Melis, Lester Mackey
Abstract: Recent breakthroughs in self-supervised Joint-Embedding Predictive Architectures (JEPAs) have established that regularizing Euclidean representations toward isotropic Gaussian priors yields provable gains in training stability and downstream generalization. We introduce a new, flexible family of KerJEPAs, self-supervised learning algorithms with kernel-based regularizers. One instance of this family corresponds to the recently-introduced LeJEPA Epps-Pulley regularizer which approximates a sliced maximum mean discrepancy (MMD) with a Gaussian prior and Gaussian kernel. By expanding the class of viable kernels and priors and computing the closed-form high-dimensional limit of sliced MMDs, we develop alternative KerJEPAs with a number of favorable properties including improved training stability and design flexibility.
Authors: Jiaqi Peng, Wenzhe Cai, Yuqiang Yang, Tai Wang, Yuan Shen, Jiangmiao Pang
Abstract: Trajectory planning in unstructured environments is a fundamental and challenging capability for mobile robots. Traditional modular pipelines suffer from latency and cascading errors across perception, localization, mapping, and planning modules. Recent end-to-end learning methods map raw visual observations directly to control signals or trajectories, promising greater performance and efficiency in open-world settings. However, most prior end-to-end approaches still rely on separate localization modules that depend on accurate sensor extrinsic calibration for self-state estimation, thereby limiting generalization across embodiments and environments. We introduce LoGoPlanner, a localization-grounded, end-to-end navigation framework that addresses these limitations by: (1) finetuning a long-horizon visual-geometry backbone to ground predictions with absolute metric scale, thereby providing implicit state estimation for accurate localization; (2) reconstructing surrounding scene geometry from historical observations to supply dense, fine-grained environmental awareness for reliable obstacle avoidance; and (3) conditioning the policy on implicit geometry bootstrapped by the aforementioned auxiliary tasks, thereby reducing error propagation.We evaluate LoGoPlanner in both simulation and real-world settings, where its fully end-to-end design reduces cumulative error while metric-aware geometry memory enhances planning consistency and obstacle avoidance, leading to more than a 27.3\% improvement over oracle-localization baselines and strong generalization across embodiments and environments. The code and models have been made publicly available on the \href{https://steinate.github.io/logoplanner.github.io/}{project page}.
Authors: Niclas Griesshaber, Jochen Streb
Abstract: We leverage multimodal large language models (LLMs) to construct a dataset of 306,070 German patents (1877-1918) from 9,562 archival image scans using our LLM-based pipeline powered by Gemini-2.5-Pro and Gemini-2.5-Flash-Lite. Our benchmarking exercise provides tentative evidence that multimodal LLMs can create higher quality datasets than our research assistants, while also being more than 795 times faster and 205 times cheaper in constructing the patent dataset from our image corpus. About 20 to 50 patent entries are embedded on each page, arranged in a double-column format and printed in Gothic and Roman fonts. The font and layout complexity of our primary source material suggests to us that multimodal LLMs are a paradigm shift in how datasets are constructed in economic history. We open-source our benchmarking and patent datasets as well as our LLM-based data pipeline, which can be easily adapted to other image corpora using LLM-assisted coding tools, lowering the barriers for less technical researchers. Finally, we explain the economics of deploying LLMs for historical dataset construction and conclude by speculating on the potential implications for the field of economic history.
Authors: Apoorv Vyas, Heng-Jui Chang, Cheng-Fu Yang, Po-Yao Huang, Luya Gao, Julius Richter, Sanyuan Chen, Matt Le, Piotr Doll\'ar, Christoph Feichtenhofer, Ann Lee, Wei-Ning Hsu
Abstract: We introduce Perception Encoder Audiovisual, PE-AV, a new family of encoders for audio and video understanding trained with scaled contrastive learning. Built on PE, PE-AV makes several key contributions to extend representations to audio, and natively support joint embeddings across audio-video, audio-text, and video-text modalities. PE-AV's unified cross-modal embeddings enable novel tasks such as speech retrieval, and set a new state of the art across standard audio and video benchmarks. We unlock this by building a strong audiovisual data engine that synthesizes high-quality captions for O(100M) audio-video pairs, enabling large-scale supervision consistent across modalities. Our audio data includes speech, music, and general sound effects-avoiding single-domain limitations common in prior work. We exploit ten pairwise contrastive objectives, showing that scaling cross-modality and caption-type pairs strengthens alignment and improves zero-shot performance. We further develop PE-A-Frame by fine-tuning PE-AV with frame-level contrastive objectives, enabling fine-grained audio-frame-to-text alignment for tasks such as sound event detection.
Authors: Yusuke Hosoya, Masanori Suganuma, Takayuki Okatani
Abstract: Open-set object detection (OSOD), a task involving the detection of unknown objects while accurately detecting known objects, has recently gained attention. However, we identify a fundamental issue with the problem formulation employed in current OSOD studies. Inherent to object detection is knowing "what to detect," which contradicts the idea of identifying "unknown" objects. This sets OSOD apart from open-set recognition (OSR). This contradiction complicates a proper evaluation of methods' performance, a fact that previous studies have overlooked. Next, we propose a novel formulation wherein detectors are required to detect both known and unknown classes within specified super-classes of object classes. This new formulation is free from the aforementioned issues and has practical applications. Finally, we design benchmark tests utilizing existing datasets and report the experimental evaluation of existing OSOD methods. The results show that existing methods fail to accurately detect unknown objects due to misclassification of known and unknown classes rather than incorrect bounding box prediction. As a byproduct, we introduce a taxonomy of OSOD, resolving confusion prevalent in the literature. We anticipate that our study will encourage the research community to reconsider OSOD and facilitate progress in the right direction.
Authors: Siming Yan, Min Bai, Weifeng Chen, Xiong Zhou, Qixing Huang, Li Erran Li
Abstract: By combining natural language understanding, generation capabilities, and breadth of knowledge of large language models with image perception, recent large vision language models (LVLMs) have shown unprecedented visual reasoning capabilities. However, the generated text often suffers from inaccurate grounding in the visual input, resulting in errors such as hallucination of nonexistent scene elements, missing significant parts of the scene, and inferring incorrect attributes of and relationships between objects. To address these issues, we introduce a novel framework, ViGoR (Visual Grounding Through Fine-Grained Reward Modeling) that utilizes fine-grained reward modeling to significantly enhance the visual grounding of LVLMs over pre-trained baselines. This improvement is efficiently achieved using much cheaper human evaluations instead of full supervisions, as well as automated methods. We show the effectiveness of our approach through a variety of evaluation methods and benchmarks. Additionally, we released our human annotation (https://github.com/amazon-science/vigor) comprising 15,440 images and generated text pairs with fine-grained evaluations to contribute to related research in the community.
Authors: Hao Fang, Jiawei Kong, Wenbo Yu, Bin Chen, Jiawei Li, Hao Wu, Shutao Xia, Ke Xu
Abstract: Vision-Language Pre-training (VLP) models have exhibited unprecedented capability in many applications by taking full advantage of the multimodal alignment. However, previous studies have shown they are vulnerable to maliciously crafted adversarial samples. Despite recent success, these methods are generally instance-specific and require generating perturbations for each input sample. In this paper, we reveal that VLP models are also vulnerable to the instance-agnostic universal adversarial perturbation (UAP). Specifically, we design a novel Contrastive-training Perturbation Generator with Cross-modal conditions (C-PGC) to achieve the attack. In light that the pivotal multimodal alignment is achieved through the advanced contrastive learning technique, we devise to turn this powerful weapon against themselves, i.e., employ a malicious version of contrastive learning to train the C-PGC based on our carefully crafted positive and negative image-text pairs for essentially destroying the alignment relationship learned by VLP models. Besides, C-PGC fully utilizes the characteristics of Vision-and-Language (V+L) scenarios by incorporating both unimodal and cross-modal information as effective guidance. Extensive experiments show that C-PGC successfully forces adversarial samples to move away from their original area in the VLP model's feature space, thus essentially enhancing attacks across various victim models and V+L tasks. The GitHub repository is available at https://github.com/ffhibnese/CPGC_VLP_Universal_Attacks.
URLs: https://github.com/ffhibnese/CPGC_VLP_Universal_Attacks.
Authors: Sunoh Lee, Minsik Jeon, Jihong Min, Junwon Seo
Abstract: Open World Object Detection(OWOD) addresses realistic scenarios where unseen object classes emerge, enabling detectors trained on known classes to detect unknown objects and incrementally incorporate the knowledge they provide. While existing OWOD methods primarily focus on detecting unknown objects, they often overlook the rich semantic relationships between detected objects, which are essential for scene understanding and applications in open-world environments (e.g., open-world tracking and novel class discovery). In this paper, we extend the OWOD framework to jointly detect unknown objects and learn semantically rich instance embeddings, enabling the detector to capture fine-grained semantic relationships between instances. To this end, we propose two modules that leverage the rich and generalizable knowledge of Vision Foundation Models(VFMs) and can be integrated into open-world object detectors. First, the Unknown Box Refine Module uses instance masks from the Segment Anything Model to accurately localize unknown objects. The Embedding Transfer Module then distills instance-wise semantic similarities from VFM features to the detector's embeddings via a relaxed contrastive loss, enabling the detector to learn a semantically meaningful and generalizable instance feature. Extensive experiments show that our method significantly improves both unknown object detection and instance embedding quality, while also enhancing performance in downstream tasks such as open-world tracking.
Authors: Amir Kazemi, Qurat ul ain Fatima, Volodymyr Kindratenko, Christopher Tessum
Abstract: Image labeling is a critical bottleneck in the development of computer vision technologies, often constraining the potential of machine learning models due to the time-intensive nature of manual annotations. This work introduces a novel approach that leverages outpainting to mitigate the problem of annotated data scarcity by generating artificial contexts and annotations, significantly reducing manual labeling efforts. We apply this technique to a particularly acute challenge in autonomous driving, urban planning, and environmental monitoring: the lack of diverse, eye-level vehicle images in desired classes. Our dataset comprises AI-generated vehicle images obtained by detecting and cropping vehicles from manually selected seed images, which are then outpainted onto larger canvases to simulate varied real-world conditions. The outpainted images include detailed annotations, providing high-quality ground truth data. Advanced outpainting techniques and image quality assessments ensure visual fidelity and contextual relevance. Ablation results show that incorporating AIDOVECL improves overall detection performance by up to 10%, and delivers gains of up to 40% in settings with greater diversity of context, object scale, and placement, with underrepresented classes achieving up to 50% higher true positives. AIDOVECL enhances vehicle detection by augmenting real training data and supporting evaluation across diverse scenarios. By demonstrating outpainting as an automatic annotation paradigm, it offers a practical and versatile solution for building fine-grained datasets with reduced labeling effort across multiple machine learning domains. The code and links to datasets used in this study are available for further research and replication at https://github.com/amir-kazemi/aidovecl .
Authors: Zaira Manigrasso, Matteo Dunnhofer, Antonino Furnari, Moritz Nottebaum, Antonio Finocchiaro, Davide Marana, Rosario Forte, Giovanni Maria Farinella, Christian Micheloni
Abstract: Episodic memory retrieval enables wearable cameras to recall objects or events previously observed in video. However, existing formulations assume an "offline" setting with full video access at query time, limiting their applicability in real-world scenarios with power and storage-constrained wearable devices. Towards more application-ready episodic memory systems, we introduce Online Visual Query 2D (OVQ2D), a task where models process video streams online, observing each frame only once, and retrieve object localizations using a compact memory instead of full video history. We address OVQ2D with ESOM (Egocentric Streaming Object Memory), a novel framework integrating an object discovery module, an object tracking module, and a memory module that find, track, and store spatio-temporal object information for efficient querying. Experiments on Ego4D demonstrate ESOM's superiority over other online approaches, though OVQ2D remains challenging, with top performance at only ~4% success. ESOM's accuracy increases markedly with perfect object tracking (31.91%), discovery (40.55%), or both (81.92%), underscoring the need of applied research on these components.
Authors: Yang Zhang, Er Jin, Wenzhong Liang, Yanfei Dong, Ashkan Khakzar, Philip Torr, Johannes Stegmaier, Kenji Kawaguchi
Abstract: Diffusion models have achieved impressive advancements in various vision tasks. However, these gains often rely on increasing model size, which escalates computational complexity and memory demands, complicating deployment, raising inference costs, and causing environmental impact. While some studies have explored pruning techniques to improve the memory efficiency of diffusion models, most existing methods require extensive retraining to retain the model performance. Retraining a modern large diffusion model is extremely costly and resource-intensive, which limits the practicality of these methods. In this work, we achieve low-cost diffusion pruning without retraining by proposing a model-agnostic structural pruning framework for diffusion models that learns a differentiable mask to sparsify the model. To ensure effective pruning that preserves the quality of the final denoised latent, we design a novel end-to-end pruning objective that spans the entire diffusion process. As end-to-end pruning is memory-intensive, we further propose time step gradient checkpointing, a technique that significantly reduces memory usage during optimization, enabling end-to-end pruning within a limited memory budget. Results on state-of-the-art U-Net diffusion models SDXL and diffusion transformers (FLUX) demonstrate that our method can effectively prune up to 20% parameters with minimal perceptible performance degradation, and notably, without the need for model retraining. We also showcase that our method can still prune on top of time step distilled diffusion models.
Authors: Zehao Wang, Xinpeng Liu, Yudonglin Zhang, Xiaoqian Wu, Zhou Fang, Yifan Fang, Junfu Pu, Cewu Lu, Yong-Lu Li
Abstract: Multimodal Large Language Models (MLLMs) have garnered significant attention recently and demonstrate outstanding capabilities in various tasks such as OCR, VQA, captioning, $\textit{etc}$. However, hallucination remains a persistent issue. While numerous methods have been proposed to mitigate hallucinations, achieving notable improvements, these methods primarily focus on mitigating hallucinations about $\textbf{object/noun-related}$ concepts. Verb concepts, crucial for understanding human actions, have been largely overlooked. In this paper, to the best of our knowledge, we are the $\textbf{first}$ to investigate the $\textbf{verb hallucination}$ phenomenon of MLLMs from various perspectives. Our findings reveal that most state-of-the-art MLLMs suffer from severe verb hallucination. To assess the effectiveness of existing mitigation methods for object concept hallucination on verb hallucination, we evaluated these methods and found that they do not effectively address verb hallucination. To address this issue, we propose a novel rich verb knowledge-based tuning method to mitigate verb hallucination. The experiment results demonstrate that our method significantly reduces hallucinations related to verbs.
Authors: Hee Jae Kim, Kathakoli Sengupta, Masaki Kuribayashi, Hernisa Kacorri, Eshed Ohn-Bar
Abstract: People who are blind perceive the world differently than those who are sighted, which can result in distinct motion characteristics. For instance, when crossing at an intersection, blind individuals may have different patterns of movement, such as veering more from a straight path or using touch-based exploration around curbs and obstacles. These behaviors may appear less predictable to motion models embedded in technologies such as autonomous vehicles. Yet, the ability of 3D motion models to capture such behavior has not been previously studied, as existing datasets for 3D human motion currently lack diversity and are biased toward people who are sighted. In this work, we introduce BlindWays, the first multimodal motion benchmark for pedestrians who are blind. We collect 3D motion data using wearable sensors with 11 blind participants navigating eight different routes in a real-world urban setting. Additionally, we provide rich textual descriptions that capture the distinctive movement characteristics of blind pedestrians and their interactions with both the navigation aid (e.g., a white cane or a guide dog) and the environment. We benchmark state-of-the-art 3D human prediction models, finding poor performance with off-the-shelf and pre-training-based methods for our novel task. To contribute toward safer and more reliable systems that can seamlessly reason over diverse human movements in their environments, our text-and-motion benchmark is available at https://blindways.github.io.
Authors: Jaehwan Jeong, Sumin In, Sieun Kim, Hannie Shin, Jongheon Jeong, Sang Ho Yoon, Jaewook Chung, Sangpil Kim
Abstract: The rising use of deepfakes in criminal activities presents a significant issue, inciting widespread controversy. While numerous studies have tackled this problem, most primarily focus on deepfake detection. These reactive solutions are insufficient as a fundamental approach for crimes where authenticity is disregarded. Existing proactive defenses also have limitations, as they are effective only for deepfake models based on specific Generative Adversarial Networks (GANs), making them less applicable in light of recent advancements in diffusion-based models. In this paper, we propose a proactive defense method named FaceShield, which introduces novel defense strategies targeting deepfakes generated by Diffusion Models (DMs) and facilitates defenses on various existing GAN-based deepfake models through facial feature extractor manipulations. Our approach consists of three main components: (i) manipulating the attention mechanism of DMs to exclude protected facial features during the denoising process, (ii) targeting prominent facial feature extraction models to enhance the robustness of our adversarial perturbation, and (iii) employing Gaussian blur and low-pass filtering techniques to improve imperceptibility while enhancing robustness against JPEG compression. Experimental results on the CelebA-HQ and VGGFace2-HQ datasets demonstrate that our method achieves state-of-the-art performance against the latest deepfake models based on DMs, while also exhibiting transferability to GANs and showcasing greater imperceptibility of noise along with enhanced robustness. Code is available here: https://github.com/kuai-lab/iccv25_faceshield
Authors: Chuang Yu, Jinmiao Zhao, Yunpeng Liu, Sicheng Zhao, Yimian Dai, Xiangyu Yue
Abstract: Recently, single-frame infrared small target (SIRST) detection with single point supervision has drawn wide-spread attention. However, the latest label evolution with single point supervision (LESPS) framework suffers from instability, excessive label evolution, and difficulty in exerting embedded network performance. Inspired by organisms gradually adapting to their environment and continuously accumulating knowledge, we construct an innovative Progressive Active Learning (PAL) framework, which drives the existing SIRST detection networks progressively and actively recognizes and learns harder samples. Specifically, to avoid the early low-performance model leading to the wrong selection of hard samples, we propose a model pre-start concept, which focuses on automatically selecting a portion of easy samples and helping the model have basic task-specific learning capabilities. Meanwhile, we propose a refined dual-update strategy, which can promote reasonable learning of harder samples and continuous refinement of pseudo-labels. In addition, to alleviate the risk of excessive label evolution, a decay factor is reasonably introduced, which helps to achieve a dynamic balance between the expansion and contraction of target annotations. Extensive experiments show that existing SIRST detection networks equipped with our PAL framework have achieved state-of-the-art (SOTA) results on multiple public datasets. Furthermore, our PAL framework can build an efficient and stable bridge between full supervision and single point supervision tasks. Our code is available at https://github.com/YuChuang1205/PAL
Authors: Qianpu Sun, Changyong Shu, Sifan Zhou, Runxi Cheng, Yongxian Wei, Zichen Yu, Dawei Yang, Sirui Han, Yuan Chun
Abstract: Weakly-supervised 3D occupancy perception is crucial for vision-based autonomous driving in outdoor environments. Previous methods based on NeRF often face a challenge in balancing the number of samples used. Too many samples can decrease efficiency, while too few can compromise accuracy, leading to variations in the mean Intersection over Union (mIoU) by 5-10 points. Furthermore, even with surrounding-view image inputs, only a single image is rendered from each viewpoint at any given moment. This limitation leads to duplicated predictions, which significantly impacts the practicality of the approach. However, this issue has largely been overlooked in existing research. To address this, we propose GSRender, which uses 3D Gaussian Splatting for weakly-supervised occupancy estimation, simplifying the sampling process. Additionally, we introduce the Ray Compensation module, which reduces duplicated predictions by compensating for features from adjacent frames. Finally, we redesign the dynamic loss to remove the influence of dynamic objects from adjacent frames. Extensive experiments show that our approach achieves SOTA results in RayIoU (+6.0), while also narrowing the gap with 3D- supervised methods. This work lays a solid foundation for weakly-supervised occupancy perception. The code is available at https://github.com/Jasper-sudo-Sun/GSRender.
Authors: Honglin Cao, Zijian Zhou, Wenjie Wei, Ammar Belatreche, Yu Liang, Dehao Zhang, Malu Zhang, Yang Yang, Haizhou Li
Abstract: Transformer-based Spiking Neural Networks (SNNs) introduce a novel event-driven self-attention paradigm that combines the high performance of Transformers with the energy efficiency of SNNs. However, the larger model size and increased computational demands of the Transformer structure limit their practicality in resource-constrained scenarios. In this paper, we integrate binarization techniques into Transformer-based SNNs and propose the Binary Event-Driven Spiking Transformer, i.e. BESTformer. The proposed BESTformer can significantly reduce storage and computational demands by representing weights and attention maps with a mere 1-bit. However, BESTformer suffers from a severe performance drop from its full-precision counterpart due to the limited representation capability of binarization. To address this issue, we propose a Coupled Information Enhancement (CIE) method, which consists of a reversible framework and information enhancement distillation. By maximizing the mutual information between the binary model and its full-precision counterpart, the CIE method effectively mitigates the performance degradation of the BESTformer. Extensive experiments on static and neuromorphic datasets demonstrate that our method achieves superior performance to other binary SNNs, showcasing its potential as a compact yet high-performance model for resource-limited edge devices. The repository of this paper is available at https://github.com/CaoHLin/BESTFormer.
Authors: Lin Chen, Qi Yang, Kun Ding, Zhihao Li, Gang Shen, Fei Li, Qiyuan Cao, Shiming Xiang
Abstract: Open-vocabulary semantic segmentation (OVSS) is an open-world task that aims to assign each pixel within an image to a specific class defined by arbitrary text descriptions. While large-scale vision-language models have shown remarkable open-vocabulary capabilities, their image-level pretraining limits effectiveness on pixel-wise dense prediction tasks like OVSS. Recent cost-based methods narrow this granularity gap by constructing pixel-text cost maps and refining them via cost aggregation mechanisms. Despite achieving promising performance, these approaches suffer from high computational costs and long inference latency. In this paper, we identify two major sources of redundancy in the cost-based OVSS framework: redundant information introduced during cost maps construction and inefficient sequence modeling in cost aggregation. To address these issues, we propose ERR-Seg, an efficient architecture that incorporates Redundancy-Reduced Hierarchical Cost maps (RRHC) and Redundancy-Reduced Cost Aggregation (RRCA). Specifically, RRHC reduces redundant class channels by customizing a compact class vocabulary for each image and integrates hierarchical cost maps to enrich semantic representation. RRCA alleviates computational burden by performing both spatial-level and class-level sequence reduction before aggregation. Overall, ERR-Seg results in a lightweight structure for OVSS, characterized by substantial memory and computational savings without compromising accuracy. Compared to previous state-of-the-art methods on the ADE20K-847 benchmark, ERR-Seg improves performance by $5.6\%$ while achieving a 3.1$\times$ speedup.
Authors: Yicen Li, Haitz S\'aez de Oc\'ariz Borde, Anastasis Kratsios, Paul D. McNicholas
Abstract: In the era of pre-trained models, effective classification can often be achieved using simple linear probing or lightweight readout layers. In contrast, many competitive clustering pipelines have a multi-modal design, leveraging large language models (LLMs) or other text encoders, and text-image pairs, which are often unavailable in real-world downstream applications. Additionally, such frameworks are generally complicated to train and require substantial computational resources, making widespread adoption challenging. In this work, we show that in deep clustering, competitive performance with more complex state-of-the-art methods can be achieved using a text-free and highly simplified training pipeline. In particular, our approach, Simple Clustering via Pre-trained models (SCP), trains only a small cluster head while leveraging pre-trained vision model feature representations and positive data pairs. Experiments on benchmark datasets, including CIFAR-10, CIFAR-20, CIFAR-100, STL-10, ImageNet-10, and ImageNet-Dogs, demonstrate that SCP achieves highly competitive performance. Furthermore, we provide a theoretical result explaining why, at least under ideal conditions, additional text-based embeddings may not be necessary to achieve strong clustering performance in vision.
Authors: Naichuan Zheng, Hailun Xia, Zeyu Liang, Yuchen Du
Abstract: Multimodal human action recognition based on RGB and skeleton data fusion, while effective, is constrained by significant limitations such as high computational complexity, excessive memory consumption, and substantial energy demands, particularly when implemented with Artificial Neural Networks (ANN). These limitations restrict its applicability in resource-constrained scenarios. To address these challenges, we propose a novel Spiking Neural Network (SNN)-driven framework for multimodal human action recognition, utilizing event camera and skeleton data. Our framework is centered on two key innovations: (1) a novel multimodal SNN architecture that employs distinct backbone networks for each modality-an SNN-based Mamba for event camera data and a Spiking Graph Convolutional Network (SGN) for skeleton data-combined with a spiking semantic extraction module to capture deep semantic representations; and (2) a pioneering SNN-based discretized information bottleneck mechanism for modality fusion, which effectively balances the preservation of modality-specific semantics with efficient information compression. To validate our approach, we propose a novel method for constructing a multimodal dataset that integrates event camera and skeleton data, enabling comprehensive evaluation. Extensive experiments demonstrate that our method achieves superior performance in both recognition accuracy and energy efficiency, offering a promising solution for practical applications.
Authors: Yi Wang, Mushui Liu, Wanggui He, Hanyang Yuan, Longxiang Zhang, Ziwei Huang, Guanghao Zhang, Wenkai Fang, Haoze Jiang, Shengxuming Zhang, Dong She, Jinlong Liu, Weilong Dai, Mingli Song, Hao Jiang, Jie Song
Abstract: Unified generative models have shown remarkable performance in text and image generation. For image synthesis tasks, they adopt straightforward text-to-image (T2I) generation. However, direct T2I generation limits the models in handling complex compositional instructions, which frequently occur in real-world scenarios. Although this issue is vital, existing works mainly focus on improving the basic image generation capability of the models. While such improvements help to some extent, they still fail to adequately resolve the problem. Inspired by Chain of Thought (CoT) solving complex problems step by step, this work aims to introduce CoT into unified generative models to address the challenges of complex image generation that direct T2I generation cannot effectively solve, thereby endowing models with enhanced image generation ability. To achieve this, we first propose Functionality-oriented eXperts (FoXperts), an expert-parallel architecture in our model FoX, which assigns experts by function. FoXperts disentangles potential conflicts in mainstream modality-oriented designs and provides a solid foundation for CoT. When introducing CoT, the first question is how to design it for complex image generation. To this end, we emulate a human-like artistic workflow--planning, acting, reflection, and correction--and propose the Multimodal Chain of Thought (MCoT) approach, as the data involves both text and image. To address the subsequent challenge of designing an effective MCoT training paradigm, we develop a multi-task joint training scheme that equips the model with all capabilities required for each MCoT step in a disentangled manner. This paradigm avoids the difficulty of collecting consistent multi-step data tuples. Extensive experiments show that FoX consistently outperforms existing unified models on various T2I benchmarks, delivering notable improvements in complex image generation.
Authors: Zhun Mou, Bin Xia, Zhengchao Huang, Wenming Yang, Jiaya Jia
Abstract: Recent great advances in video generation models have demonstrated their potential to produce high-quality videos, bringing challenges to effective evaluation. Unlike human evaluation, existing automated evaluation metrics lack highlevel semantic understanding and reasoning capabilities for video, thus making them infeasible and unexplainable. To fill this gap, we curate GRADEO-Instruct, a multi-dimensional T2V evaluation instruction tuning dataset, including 3.3k videos from over 10 existing video generation models and multi-step reasoning assessments converted by 16k human annotations. We then introduce GRADEO, one of the first specifically designed video evaluation models, which grades AI-generated videos for explainable scores and assessments through multi-step reasoning. Experiments show that our method aligns better with human evaluations than existing methods. Furthermore, our benchmarking reveals that current video generation models struggle to produce content that aligns with human reasoning and complex real-world scenarios.
Authors: Sanghyun Jo, Ziseok Lee, Wooyeol Lee, Jonghyun Choi, Jaesik Park, Kyungsu Kim
Abstract: High-quality instance and panoptic segmentation has traditionally relied on dense instance-level annotations such as masks, boxes, or points, which are costly, inconsistent, and difficult to scale. Unsupervised and weakly-supervised approaches reduce this burden but remain constrained by semantic backbone constraints and human bias, often producing merged or fragmented outputs. We present TRACE (TRAnsforming diffusion Cues to instance Edges), showing that text-to-image diffusion models secretly function as instance edge annotators. TRACE identifies the Instance Emergence Point (IEP) where object boundaries first appear in self-attention maps, extracts boundaries through Attention Boundary Divergence (ABDiv), and distills them into a lightweight one-step edge decoder. This design removes the need for per-image diffusion inversion, achieving 81x faster inference while producing sharper and more connected boundaries. On the COCO benchmark, TRACE improves unsupervised instance segmentation by +5.1 AP, and in tag-supervised panoptic segmentation it outperforms point-supervised baselines by +1.7 PQ without using any instance-level labels. These results reveal that diffusion models encode hidden instance boundary priors, and that decoding these signals offers a practical and scalable alternative to costly manual annotation. Code is available at https://github.com/shjo-april/DiffEGG.
Authors: Boyu Chen, Zhengrong Yue, Siran Chen, Zikang Wang, Yang Liu, Peng Li, Yali Wang
Abstract: Existing MLLMs encounter significant challenges in modeling the temporal context within long videos. Currently, mainstream Agent-based methods use external tools to assist a single MLLM in answering long video questions. Despite such tool-based support, a solitary MLLM still offers only a partial understanding of long videos, resulting in limited performance. In order to better address long video tasks, we introduce LVAgent, the first framework enabling multi-round dynamic collaboration of MLLM agents in long video understanding. Our method consists of four key steps: 1) Selection: We pre-select appropriate agents from the model library to form optimal agent teams based on different tasks. 2) Perception: We design an effective retrieval scheme for long videos to improve the coverage of critical temporal segments while maintaining computational efficiency. 3) Action: Agents answer long video questions and exchange reasons. 4) Reflection: We evaluate each agent's performance in each round of discussion and optimize the agent team for dynamic collaboration. The agents iteratively refine their answers by multi-round dynamical collaboration of MLLM agents. LVAgent is the first agent system method that outperforms all closed-source models (like GPT-4o) and open-source models (like InternVL-2.5 and Qwen2-VL) in the long video understanding tasks. Our LVAgent achieves an accuracy of 80\% on four mainstream long video understanding tasks. Notably, LVAgent improves accuracy by 13.3\% on LongVideoBench. Code is available at https://github.com/64327069/LVAgent.
Authors: Hongyu Zhang, Yufan Deng, Shenghai Yuan, Yian Zhao, Peng Jin, Xuehan Hou, Chang Liu, Jie Chen
Abstract: In the domain of text-to-video (T2V) generation, reliably synthesizing compositional content involving multiple subjects with intricate relations is still underexplored. The main challenges are twofold: 1) Subject presence, where not all subjects can be presented in the video; 2) Inter-subject relations, where the interaction and spatial relationship between subjects are misaligned. Existing methods adopt techniques, such as inference-time latent optimization or layout control, which fail to address both issues simultaneously. To tackle these problems, we propose Comp-Attn, a composition-aware cross-attention variant that follows a Present-and-Align paradigm: it decouples the two challenges by enforcing subject presence at the condition level and achieving relational alignment at the attention-distribution level. Specifically, 1) We introduce Subject-aware Condition Interpolation (SCI) to reinforce subject-specific conditions and ensure each subject's presence; 2) We propose Layout-forcing Attention Modulation (LAM), which dynamically enforces the attention distribution to align with the relational layout of multiple subjects. Comp-Attn can be seamlessly integrated into various T2V baselines in a training-free manner, boosting T2V-CompBench scores by 15.7\% and 11.7\% on Wan2.1-T2V-14B and Wan2.2-T2V-A14B with only a 5\% increase in inference time. Meanwhile, it also achieves strong performance on VBench and T2I-CompBench, demonstrating its scalability in general video generation and compositional text-to-image (T2I) tasks.
Authors: Yilan Dong, Wenqing Wang, Qing Wang, Jiahao Yang, Haohe Liu, Xiatuan Zhu, Gregory Slabaugh, Shanxin Yuan
Abstract: Creating animatable hand avatars from multi-view images requires modeling complex articulations and maintaining structural consistency across poses in real time. We present HandSCS, a structure-guided 3D Gaussian Splatting framework for high-fidelity hand animation. Unlike existing approaches that condition all Gaussians on the same global pose parameters, which are inadequate for highly articulated hands, HandSCS equips each Gaussian with explicit structural guidance from both intra-pose and inter-pose perspectives. To establish intra-pose structural guidance, we introduce a Structural Coordinate Space (SCS), which bridges the gap between sparse bones and dense Gaussians through hybrid static-dynamic coordinate basis and angular-radial descriptors. To improve cross-pose coherence, we further introduce an Inter-pose Consistency Loss that promotes consistent Gaussian attributes under similar articulations. Together, these components achieve high-fidelity results with consistent fine details, even in challenging high-deformation and self-contact regions. Experiments on the InterHand2.6M dataset demonstrate that HandSCS achieves state-of-the-art performance in hand avatar animation, confirming the effectiveness of explicit structural modeling.
Authors: Kai Zhu, Li Chen, Dianshuo Li, Yunxiang Cao, Jun Cheng
Abstract: Curvilinear structure segmentation (CSS) is essential in various domains, including medical imaging, landscape analysis, industrial surface inspection, and plant analysis. While existing methods achieve high performance within specific domains, their generalizability is limited. On the other hand, large-scale models such as Segment Anything Model (SAM) exhibit strong generalization but are not optimized for curvilinear structures. Existing adaptations of SAM primarily focus on general object segmentation and lack specialized design for CSS tasks. To bridge this gap, we propose the Universal Curvilinear structure Segmentation (UCS) model, which adapts SAM to CSS tasks while further enhancing its cross-domain generalization. UCS features a novel encoder architecture integrating a pretrained SAM encoder with two innovations: a Sparse Adapter, strategically inserted to inherit the pre-trained SAM encoder's generalization capability while minimizing the number of fine-tuning parameters, and a Prompt Generation module, which leverages Fast Fourier Transform with a high-pass filter to generate curve-specific prompts. Furthermore, the UCS incorporates a mask decoder that eliminates reliance on manual interaction through a dual-compression module: a Hierarchical Feature Compression module, which aggregates the outputs of the sampled encoder to enhance detail preservation, and a Guidance Feature Compression module, which extracts and compresses image-driven guidance features. Evaluated on a comprehensive multi-domain dataset, including an in-house dataset covering eight natural curvilinear structures, UCS demonstrates state-of-the-art generalization and open-set segmentation performance across medical, engineering, natural, and plant imagery, establishing a new benchmark for universal CSS. The source code is available at https://github.com/kylechuuuuu/UCS.
Authors: Jiawei Wu, Zhifei Yang, Zhe Wang, Zhi Jin
Abstract: All-in-one image restoration (AIR) aims to address diverse degradations within a unified model by leveraging informative degradation conditions to guide the restoration process. However, existing methods often rely on implicitly learned priors, which may entangle feature representations and hinder performance in complex or unseen scenarios. Histogram of Oriented Gradients (HOG) as a classical gradient representation, we observe that it has strong discriminative capability across diverse degradations, making it a powerful and interpretable prior for AIR. Based on this insight, we propose HOGformer, a Transformer-based model that integrates learnable HOG features for degradation-aware restoration. The core of HOGformer is a Dynamic HOG-aware Self-Attention (DHOGSA) mechanism, which adaptively models long-range spatial dependencies conditioned on degradation-specific cues encoded by HOG descriptors. To further adapt the heterogeneity of degradations in AIR, we propose a Dynamic Interaction Feed-Forward (DIFF) module that facilitates channel-spatial interactions, enabling robust feature transformation under diverse degradations. Besides, we propose a HOG loss to explicitly enhance structural fidelity and edge sharpness. Extensive experiments on a variety of benchmarks, including adverse weather and natural degradations, demonstrate that HOGformer achieves state-of-the-art performance and generalizes well to complex real-world scenarios.Code is available at https://github.com/Fire-friend/HOGformer.
Authors: Sihang Chen, Lijun Yun, Ze Liu, JianFeng Zhu, Jie Chen, Hui Wang, Yueping Nie
Abstract: Deep learning techniques have achieved remarkable success in the semantic segmentation of remote sensing images and in land-use change detection. Nevertheless, their real-time deployment on edge platforms remains constrained by decoder complexity. Herein, we introduce LightFormer, a lightweight decoder for time-critical tasks that involve unstructured targets, such as disaster assessment, unmanned aerial vehicle search-and-rescue, and cultural heritage monitoring. LightFormer employs a feature-fusion and refinement module built on channel processing and a learnable gating mechanism to aggregate multi-scale, multi-range information efficiently, which drastically curtails model complexity. Furthermore, we propose a spatial information selection module (SISM) that integrates long-range attention with a detail preservation branch to capture spatial dependencies across multiple scales, thereby substantially improving the recognition of unstructured targets in complex scenes. On the ISPRS Vaihingen benchmark, LightFormer attains 99.9% of GLFFNet's mIoU (83.9% vs. 84.0%) while requiring only 14.7% of its FLOPs and 15.9% of its parameters, thus achieving an excellent accuracy-efficiency trade-off. Consistent results on LoveDA, ISPRS Potsdam, RescueNet, and FloodNet further demonstrate its robustness and superior perception of unstructured objects. These findings highlight LightFormer as a practical solution for remote sensing applications where both computational economy and high-precision segmentation are imperative.
Authors: Lvpan Cai, Haowei Wang, Jiayi Ji, Yanshu Zhoumen, Shen Chen, Taiping Yao, Xiaoshuai Sun
Abstract: The rise of AI-generated image tools has made localized forgeries increasingly realistic, posing challenges for visual content integrity. Although recent efforts have explored localized AIGC detection, existing datasets predominantly focus on object-level forgeries while overlooking broader scene edits in regions such as sky or ground. To address these limitations, we introduce \textbf{BR-Gen}, a large-scale dataset of 150,000 locally forged images with diverse scene-aware annotations, which are based on semantic calibration to ensure high-quality samples. BR-Gen is constructed through a fully automated ``Perception-Creation-Evaluation'' pipeline to ensure semantic coherence and visual realism. In addition, we further propose \textbf{NFA-ViT}, a Noise-guided Forgery Amplification Vision Transformer that enhances the detection of localized forgeries by amplifying subtle forgery-related features across the entire image. NFA-ViT mines heterogeneous regions in images, \emph{i.e.}, potential edited areas, by noise fingerprints. Subsequently, attention mechanism is introduced to compel the interaction between normal and abnormal features, thereby propagating the traces throughout the entire image, allowing subtle forgeries to influence a broader context and improving overall detection robustness. Extensive experiments demonstrate that BR-Gen constructs entirely new scenarios that are not covered by existing methods. Take a step further, NFA-ViT outperforms existing methods on BR-Gen and generalizes well across current benchmarks.
Authors: Feng Yang, Wenliang Qian, Wangmeng Zuo, Hui Li
Abstract: Score Distillation Sampling (SDS) leverages pretrained 2D diffusion models to advance text-to-3D generation but neglects multi-view correlations, being prone to geometric inconsistencies and multi-face artifacts in the generated 3D content. In this work, we propose Coupled Score Distillation (CSD), a framework that couples multi-view joint distribution priors to ensure geometrically consistent 3D generation while enabling the stable and direct optimization of 3D Gaussian Splatting. Specifically, by reformulating the optimization as a multi-view joint optimization problem, we derive an effective optimization rule that effectively couples multi-view priors to guide optimization across different viewpoints while preserving the diversity of generated 3D assets. Additionally, we propose a framework that directly optimizes 3D Gaussian Splatting (3D-GS) with random initialization to generate geometrically consistent 3D content. We further employ a deformable tetrahedral grid, initialized from 3D-GS and refined through CSD, to produce high-quality, refined meshes. Quantitative and qualitative experimental results demonstrate the efficiency and competitive quality of our approach.
Authors: Chuanzhi Xu, Haoxian Zhou, Langyi Chen, Haodong Chen, Zeke Zexi Hu, Zhicheng Lu, Ying Zhou, Vera Chung, Qiang Qu, Weidong Cai
Abstract: Event cameras are rapidly emerging as powerful vision sensors for 3D reconstruction, uniquely capable of asynchronously capturing per-pixel brightness changes. Compared to traditional frame-based cameras, event cameras produce sparse yet temporally dense data streams, enabling robust and accurate 3D reconstruction even under challenging conditions such as high-speed motion, low illumination, and extreme dynamic range scenarios. These capabilities offer substantial promise for transformative applications across various fields, including autonomous driving, robotics, aerial navigation, and immersive virtual reality. In this survey, we present the first comprehensive review exclusively dedicated to event-based 3D reconstruction. Existing approaches are systematically categorised based on input modality into stereo, monocular, and multimodal systems, and further classified according to reconstruction methodologies, including geometry-based techniques, deep learning approaches, and neural rendering techniques such as Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS). Within each category, methods are chronologically organised to highlight the evolution of key concepts and advancements. Furthermore, we provide a detailed summary of publicly available datasets specifically suited to event-based reconstruction tasks. Finally, we discuss significant open challenges in dataset availability, standardised evaluation, effective representation, and dynamic scene reconstruction, outlining insightful directions for future research. This survey aims to serve as an essential reference and provides a clear and motivating roadmap toward advancing the state of the art in event-driven 3D reconstruction.
Authors: Weilai Xiang, Hongyu Yang, Di Huang, Yunhong Wang
Abstract: While diffusion models excel at image synthesis, useful representations have been shown to emerge from generative pre-training, suggesting a path towards unified generative and discriminative learning. However, suboptimal semantic flow within current architectures can hinder this potential: features encoding the richest high-level semantics are underutilized and diluted when propagating through decoding layers, impeding the formation of an explicit semantic bottleneck layer. To address this, we introduce self-conditioning, a lightweight mechanism that reshapes the model's layer-wise semantic hierarchy without external guidance. By aggregating and rerouting intermediate features to guide subsequent decoding layers, our method concentrates more high-level semantics, concurrently strengthening global generative guidance and forming more discriminative representations. This simple approach yields a dual-improvement trend across pixel-space UNet, UViT and latent-space DiT models with minimal overhead. Crucially, it creates an architectural semantic bridge that propagates discriminative improvements into generation and accommodates further techniques such as contrastive self-distillation. Experiments show that our enhanced models, especially self-conditioned DiT, are powerful dual learners that yield strong and transferable representations on image and dense classification tasks, surpassing various generative self-supervised models in linear probing while also improving or maintaining high generation quality.
Authors: Shilin Yan, Jiaming Han, Joey Tsai, Hongwei Xue, Rongyao Fang, Lingyi Hong, Ziyu Guo, Ray Zhang
Abstract: The advent of Large Multimodal Models (LMMs) has significantly enhanced Large Language Models (LLMs) to process and interpret diverse data modalities (e.g., image and video). However, as input complexity increases, particularly with long video sequences, the number of required tokens has grown significantly, leading to quadratically computational costs. This has made the efficient compression of video tokens in LMMs, while maintaining performance integrity, a pressing research challenge. In this paper, we introduce CrossLMM, decoupling long video sequences from LMMs via a dual cross-attention mechanism, which substantially reduces visual token quantity with minimal performance degradation. Specifically, we first implement a significant token reduction from pretrained visual encoders through a pooling methodology. Then, within LLM layers, we employ a visual-to-visual cross-attention mechanism, wherein the pooled visual tokens function as queries against the original visual token set. This module enables more efficient token utilization while retaining fine-grained informational fidelity. In addition, we introduce a text-to-visual cross-attention mechanism, for which the text tokens are enhanced through interaction with the original visual tokens, enriching the visual comprehension of the text tokens. Comprehensive empirical evaluation demonstrates that our approach achieves comparable or superior performance across diverse video-based LMM benchmarks, despite utilizing substantially fewer computational resources.
Authors: Zhenhao Zhang, Ye Shi, Lingxiao Yang, Suting Ni, Qi Ye, Jingya Wang
Abstract: Understanding and synthesizing realistic 3D hand-object interactions (HOI) is critical for applications ranging from immersive AR/VR to dexterous robotics. Existing methods struggle with generalization, performing well on closed-set objects and predefined tasks but failing to handle unseen objects or open-vocabulary instructions. We introduce OpenHOI, the first framework for open-world HOI synthesis, capable of generating long-horizon manipulation sequences for novel objects guided by free-form language commands. Our approach integrates a 3D Multimodal Large Language Model (MLLM) fine-tuned for joint affordance grounding and semantic task decomposition, enabling precise localization of interaction regions (e.g., handles, buttons) and breakdown of complex instructions (e.g., "Find a water bottle and take a sip") into executable sub-tasks. To synthesize physically plausible interactions, we propose an affordance-driven diffusion model paired with a training-free physics refinement stage that minimizes penetration and optimizes affordance alignment. Evaluations across diverse scenarios demonstrate OpenHOI's superiority over state-of-the-art methods in generalizing to novel object categories, multi-stage tasks, and complex language instructions. Our project page at \href{https://openhoi.github.io}
Authors: Boyu Chen, Siran Chen, Kunchang Li, Qinglin Xu, Yu Qiao, Yali Wang
Abstract: Video understanding has been considered as one critical step towards world modeling, which is an important long-term problem in AI research. Recently, multimodal foundation models have shown such potential via large-scale pretraining. These models effectively align encoders of different modalities via contrastive learning. To further enhance performance on complex target movements and diversified video scenes, we propose to augment this alignment with deeper multimodal interactions, which are critical for understanding complex target movements with diversified video scenes. To fill this gap, we propose a unified Super Encoding Network (SEN) for video understanding, which builds up such distinct interactions through the recursive association of multimodal encoders in the foundation models. Specifically, we creatively treat those well-trained encoders as ``super neurons" in our SEN. Via designing a Recursive Association (RA) block, we progressively fuse multi-modalities with the input video, based on knowledge integrating, distributing, and prompting of super neurons in a recursive manner. In this way, our SEN can effectively encode deeper multimodal interactions for prompting various video understanding tasks in the downstream. Extensive experiments show that our SEN can remarkably boost the four most representative video tasks, including tracking, recognition, chatting, and editing, e.g., for pixel-level tracking, the average jaccard index improves 2.7%, and temporal coherence(TC) drops by 8.8% compared to the popular CaDeX++ approach. For one-shot video editing, textual alignment improves 6.4%, and frame consistency increases by 4.1% compared to the Tune-A-Video approach.
Authors: Liang Yin, Xudong Xie, Zhang Li, Xiang Bai, Yuliang Liu
Abstract: Scene text retrieval has made significant progress with the assistance of accurate text localization. However, existing approaches typically require costly bounding box annotations for training. Besides, they mostly adopt a customized retrieval strategy but struggle to unify various types of queries to meet diverse retrieval needs. To address these issues, we introduce Muti-query Scene Text retrieval with Attention Recycling (MSTAR), a box-free approach for scene text retrieval. It incorporates progressive vision embedding to dynamically capture the multi-grained representation of texts and harmonizes free-style text queries with style-aware instructions. Additionally, a multi-instance matching module is integrated to enhance vision-language alignment. Furthermore, we build the Multi-Query Text Retrieval (MQTR) dataset, the first benchmark designed to evaluate the multi-query scene text retrieval capability of models, comprising four query types and 16k images. Extensive experiments demonstrate the superiority of our method across seven public datasets and the MQTR dataset. Notably, MSTAR marginally surpasses the previous state-of-the-art model by 6.4% in MAP on Total-Text while eliminating box annotation costs. Moreover, on the MQTR benchmark, MSTAR significantly outperforms the previous models by an average of 8.5%. The code and datasets are available at https://github.com/yingift/MSTAR.
Authors: Saemee Choi, Sohyun Jeong, Hyojin Jang, Jaegul Choo, Jinhee Kim
Abstract: We propose VINO, the first zero-shot, training-free video editing method conditioned on both image and text. Our approach introduces $\rho$-start sampling and dilated dual masking to construct structured noise maps that enable coherent and accurate edits. To further enhance visual fidelity, we present zero image guidance, a controllable negative prompt strategy. Extensive experiments demonstrate that VINO faithfully incorporates the reference image into video edits, achieving strong performance compared to state-of-the-art baselines, all without any test-time or instance-specific training.
Authors: Nicolas Caytuiro, Ivan Sipiran
Abstract: Recent advances in deep learning have significantly transformed the field of 3D shape generation, enabling the synthesis of complex, diverse, and semantically meaningful 3D objects. This survey provides a comprehensive overview of the current state-of-the-art in 3D shape generation, organizing the discussion around three core components: shape representations, generative modeling approaches, and evaluation protocols. We begin by categorizing 3D representations into explicit, implicit, and hybrid setups, highlighting their structural properties, advantages, and limitations. Next, we review a wide range of generation methods, focusing on feedforward architectures. We further summarize commonly used datasets and evaluation metrics that assess fidelity, diversity, and realism of generated shapes. Finally, we identify open challenges and outline future research directions that could drive progress in controllable, efficient, and high-quality 3D shape generation. This survey aims to serve as a valuable reference for researchers and practitioners seeking a structured and in-depth understanding of this rapidly evolving field.
Authors: JaeHyuck Choi, MinJun Kim, Je Hyeong Hong
Abstract: Few-shot anomaly generation is a key challenge in industrial quality control. Although diffusion models are promising, existing methods struggle: global prompt-guided approaches corrupt normal regions, and existing inpainting-based methods often lack the in-distribution diversity essential for robust downstream models. We propose MAGIC, a fine-tuned inpainting framework that generates high-fidelity anomalies that strictly adhere to the mask while maximizing this diversity. MAGIC introduces three complementary components: (i) Gaussian prompt perturbation, which prevents model overfitting in the few-shot setting by learning and sampling from a smooth manifold of realistic anomalies, (ii) spatially adaptive guidance that applies distinct guidance strengths to the anomaly and background regions, and (iii) context-aware mask alignment to relocate masks for plausible placement within the host object. Under consistent identical evaluation protocol, MAGIC outperforms state-of-the-art methods on diverse anomaly datasets in downstream tasks
Authors: Zhanjiang Yang, Lijun Sun, Jiawei Dong, Xiaoxin An, Yang Liu, Meng Li
Abstract: Reconstructing hyperspectral images (HSIs) from RGB inputs provides a cost-effective alternative to hyperspectral cameras, but reconstructing high-dimensional spectra from three channels is inherently ill-posed. Existing methods typically directly regress RGB-to-HSI mappings using large attention networks, which are computationally expensive and handle ill-posedness only implicitly. We propose MCGA, a Mixture-of-Codebooks with Grayscale-aware Attention framework that explicitly addresses these challenges using spectral priors and photometric consistency. MCGA first learns transferable spectral priors via a mixture-of-codebooks (MoC) from heterogeneous HSI datasets, then aligns RGB features with these priors through grayscale-aware photometric attention (GANet). Efficiency and robustness are further improved via top-K attention design and test-time adaptation (TTA). Experiments on multiple real-world benchmarks demonstrate the state-of-the-art accuracy, strong cross-dataset generalization, and 4-5x faster inference. Codes will be available once acceptance at https://github.com/Fibonaccirabbit/MCGA.
Authors: Ekta Balkrishna Gavas, Sudipta Banerjee, Chinmay Hegde, Nasir Memon
Abstract: Accurate age verification can protect underage users from unauthorized access to online platforms and e-commerce sites that provide age-restricted services. However, accurate age estimation can be confounded by several factors, including facial makeup that can induce changes to alter perceived identity and age to fool both humans and machines. In this work, we propose \textsc{DiffClean} which erases makeup traces using a text-guided diffusion model to defend against makeup attacks without requiring any reference image unlike prior work. \textsc{DiffClean} improves age estimation (minor vs. adult accuracy by 5.8\%) and face verification (TMR by 5.1\% at FMR=0.01\%) compared to images with makeup. Our method is: (1) robust across digitally simulated and real-world makeup styles with high visual fidelity, (2) can be easily integrated as a pre-processing module in existing age and identity verification frameworks, and (3) advances the state-of-the art in terms of biometric and perceptual utility. Our codes are available at https://github.com/Ektagavas/DiffClean
Authors: Chihiro Noguchi, Takaki Yamamoto
Abstract: Accurate perception of the surrounding environment is essential for safe autonomous driving. 3D occupancy prediction, which estimates detailed 3D structures of roads, buildings, and other objects, is particularly important for vision-centric autonomous driving systems that do not rely on LiDAR sensors. However, in 3D semantic occupancy prediction -- where each voxel is assigned a semantic label -- annotated LiDAR point clouds are required, making data acquisition costly. In contrast, large-scale binary occupancy data, which only indicate occupied or free space without semantic labels, can be collected at a lower cost. Despite their availability, the potential of leveraging such data remains unexplored. In this study, we investigate the utilization of large-scale binary occupancy data from two perspectives: (1) pre-training and (2) learning-based auto-labeling. We propose a novel binary occupancy-based framework that decomposes the prediction process into binary and semantic occupancy modules, enabling effective use of binary occupancy data. Our experimental results demonstrate that the proposed framework outperforms existing methods in both pre-training and auto-labeling tasks, highlighting its effectiveness in enhancing 3D semantic occupancy prediction. The code is available at https://github.com/ToyotaInfoTech/b2s-occupancy
Authors: Jiahui Zhang, Yuelei Li, Anpei Chen, Muyu Xu, Kunhao Liu, Jianyuan Wang, Xiao-Xiao Long, Hanxue Liang, Zexiang Xu, Hao Su, Christian Theobalt, Christian Rupprecht, Andrea Vedaldi, Kaichen Zhou, Hanspeter Pfister, Paul Pu Liang, Shijian Lu, Fangneng Zhan
Abstract: 3D reconstruction and view synthesis are foundational problems in computer vision, graphics, and immersive technologies such as augmented reality (AR), virtual reality (VR), and digital twins. Traditional methods rely on computationally intensive iterative optimization in a complex chain, limiting their applicability in real-world scenarios. Recent advances in feed-forward approaches, driven by deep learning, have revolutionized this field by enabling fast and generalizable 3D reconstruction and view synthesis. This survey offers a comprehensive review of feed-forward techniques for 3D reconstruction and view synthesis, with a taxonomy according to the underlying representation architectures including point cloud, 3D Gaussian Splatting (3DGS), Neural Radiance Fields (NeRF), etc. We examine key tasks such as pose-free reconstruction, dynamic 3D reconstruction, and 3D-aware image and video synthesis, highlighting their applications in digital humans, SLAM, robotics, and beyond. In addition, we review commonly used datasets with detailed statistics, along with evaluation protocols for various downstream tasks. We conclude by discussing open research challenges and promising directions for future work, emphasizing the potential of feed-forward approaches to advance the state of the art in 3D vision.
Authors: Ko Watanabe, Stanislav Frolov, Aya Hassan, David Dembinsky, Adriano Lucieri, Andreas Dengel
Abstract: Recent advances in deep learning and on-device inference could transform routine screening for skin cancers. Along with the anticipated benefits of this technology, potential dangers arise from unforeseen and inherent biases. A significant obstacle is building evaluation datasets that accurately reflect key demographics, including sex, age, and race, as well as other underrepresented groups. To address this, we train a state-of-the-art generative model to generate synthetic data in a controllable manner to assess the fairness of publicly available skin cancer classifiers. To evaluate whether synthetic images can be used as a fairness testing dataset, we prepare a real-image dataset (MILK10K) as a benchmark and compare the True Positive Rate result of three models (DeepGuide, MelaNet, and SkinLesionDensnet). As a result, the classification tendencies observed in each model when tested on real and generated images showed similar patterns across different attribute data sets. We confirm that highly realistic synthetic images facilitate model fairness verification.
Authors: Kaiyue Zhou, Zelong Tan, Hongxiao Wang, Ya-Li Li, Shengjin Wang
Abstract: Recovering point clouds involves the sequential process of sampling and restoration, yet existing methods struggle to effectively leverage both topological and geometric attributes. To address this, we propose an end-to-end architecture named \textbf{TopGeoFormer}, which maintains these critical properties throughout the sampling and restoration phases. First, we revisit traditional feature extraction techniques to yield topological embedding using a continuous mapping of relative relationships between neighboring points, and integrate it in both phases for preserving the structure of the original space. Second, we propose the \textbf{InterTwining Attention} to fully merge topological and geometric embeddings, which queries shape with local awareness in both phases to form a learnable 3D shape context facilitated with point-wise, point-shape-wise, and intra-shape features. Third, we introduce a full geometry loss and a topological constraint loss to optimize the embeddings in both Euclidean and topological spaces. The geometry loss uses inconsistent matching between coarse-to-fine generations and targets for reconstructing better geometric details, and the constraint loss limits embedding variances for better approximation of the topological space. In experiments, we comprehensively analyze the circumstances using the conventional and learning-based sampling/upsampling/recovery algorithms. The quantitative and qualitative results demonstrate that our method significantly outperforms existing sampling and recovery methods.
Authors: Alfie Roddan, Tobias Czempiel, Chi Xu, Daniel S. Elson, Stamatia Giannarou
Abstract: We present SAMSA 2.0, an interactive segmentation framework for hyperspectral medical imaging that introduces spectral angle prompting to guide the Segment Anything Model (SAM) using spectral similarity alongside spatial cues. This early fusion of spectral information enables more accurate and robust segmentation across diverse spectral datasets. Without retraining, SAMSA 2.0 achieves up to +3.8% higher Dice scores compared to RGB-only models and up to +3.1% over prior spectral fusion methods. Our approach enhances few-shot and zero-shot performance, demonstrating strong generalization in challenging low-data and noisy scenarios common in clinical imaging.
Authors: Ranran Huang, Krystian Mikolajczyk
Abstract: We introduce SPFSplat, an efficient framework for 3D Gaussian splatting from sparse multi-view images, requiring no ground-truth poses during training or inference. It employs a shared feature extraction backbone, enabling simultaneous prediction of 3D Gaussian primitives and camera poses in a canonical space from unposed inputs within a single feed-forward step. Alongside the rendering loss based on estimated novel-view poses, a reprojection loss is integrated to enforce the learning of pixel-aligned Gaussian primitives for enhanced geometric constraints. This pose-free training paradigm and efficient one-step feed-forward design make SPFSplat well-suited for practical applications. Remarkably, despite the absence of pose supervision, SPFSplat achieves state-of-the-art performance in novel view synthesis even under significant viewpoint changes and limited image overlap. It also surpasses recent methods trained with geometry priors in relative pose estimation. Code and trained models are available on our project page: https://ranrhuang.github.io/spfsplat/.
Authors: Taha Mustapha Nehdi, Nairouz Mrabah, Atif Belal, Marco Pedersoli, Eric Granger
Abstract: Adapting person re-identification (reID) models to new target environments remains a challenging problem that is typically addressed using unsupervised domain adaptation (UDA) methods. Recent works show that when labeled data originates from several distinct sources (e.g., datasets and cameras), considering each source separately and applying multi-source domain adaptation (MSDA) typically yields higher accuracy and robustness compared to blending the sources and performing conventional UDA. However, state-of-the-art MSDA methods learn domain-specific backbone models or require access to source domain data during adaptation, resulting in significant growth in training parameters and computational cost. In this paper, a Source-free Adaptive Gated Experts (SAGE-reID) method is introduced for person reID. Our SAGE-reID is a cost-effective, source-free MSDA method that first trains individual source-specific low-rank adapters (LoRA) through source-free UDA. Next, a lightweight gating network is introduced and trained to dynamically assign optimal merging weights for fusion of LoRA experts, enabling effective cross-domain knowledge transfer. While the number of backbone parameters remains constant across source domains, LoRA experts scale linearly but remain negligible in size (<= 2% of the backbone), reducing both the memory consumption and risk of overfitting. Extensive experiments conducted on three challenging benchmarks: Market-1501, DukeMTMC-reID, and MSMT17 indicate that SAGE-reID outperforms state-of-the-art methods while being computationally efficient.
Authors: Qingwen Zhang, Xiaomeng Zhu, Yushan Zhang, Yixi Cai, Olov Andersson, Patric Jensfelt
Abstract: Previous dominant methods for scene flow estimation focus mainly on input from two consecutive frames, neglecting valuable information in the temporal domain. While recent trends shift towards multi-frame reasoning, they suffer from rapidly escalating computational costs as the number of frames grows. To leverage temporal information more efficiently, we propose DeltaFlow ($\Delta$Flow), a lightweight 3D framework that captures motion cues via a $\Delta$ scheme, extracting temporal features with minimal computational cost, regardless of the number of frames. Additionally, scene flow estimation faces challenges such as imbalanced object class distributions and motion inconsistency. To tackle these issues, we introduce a Category-Balanced Loss to enhance learning across underrepresented classes and an Instance Consistency Loss to enforce coherent object motion, improving flow accuracy. Extensive evaluations on the Argoverse 2, Waymo and nuScenes datasets show that $\Delta$Flow achieves state-of-the-art performance with up to 22% lower error and $2\times$ faster inference compared to the next-best multi-frame supervised method, while also demonstrating a strong cross-domain generalization ability. The code is open-sourced at https://github.com/Kin-Zhang/DeltaFlow along with trained model weights.
Authors: Stefanos Pasios, Nikos Nikolaidis
Abstract: Photorealism is an important aspect of modern video games since it can shape player experience and impact immersion, narrative engagement, and visual fidelity. To achieve photorealism, beyond traditional rendering pipelines, generative models have been increasingly adopted as an effective approach for bridging the gap between the visual realism of synthetic and real worlds. However, under real-time constraints of video games, existing generative approaches continue to face a tradeoff between visual quality and runtime efficiency. In this work, we present a framework for enhancing the photorealism of rendered game frames using generative networks. We propose REGEN, which first employs a robust unpaired image-to-image translation model to generate semantically consistent photorealistic frames. These generated frames are then used to create a paired dataset, which transforms the problem to a simpler unpaired image-to-image translation. This enables training with a lightweight method, achieving real-time inference without compromising visual quality. We evaluate REGEN on Unreal Engine, showing, by employing the CMMD metric, that it achieves comparable or slightly improved visual quality compared to the robust method, while improving the frame rate by 12x. Additional experiments also validate that REGEN adheres to the semantic preservation of the initial robust image-to-image translation method and maintains temporal consistency. Code, pre-trained models, and demos for this work are available at: https://github.com/stefanos50/REGEN
Authors: Zhixuan Liang, Yizhuo Li, Tianshuo Yang, Chengyue Wu, Sitong Mao, Tian Nian, Liuao Pei, Shunbo Zhou, Xiaokang Yang, Jiangmiao Pang, Yao Mu, Ping Luo
Abstract: Vision-Language-Action (VLA) models adapt large vision-language backbones to map images and instructions into robot actions. However, prevailing VLAs either generate actions auto-regressively in a fixed left-to-right order or attach separate MLP or diffusion heads outside the backbone, leading to fragmented information pathways and specialized training requirements that hinder a unified, scalable architecture. We present Discrete Diffusion VLA, a unified-transformer policy that models discretized action chunks with discrete diffusion. The design retains diffusion's progressive refinement paradigm while remaining natively compatible with the discrete token interface of VLMs. Our method achieves an adaptive decoding order that resolves easy action elements before harder ones and uses secondary re-masking to revisit uncertain predictions across refinement rounds, which improves consistency and enables robust error correction. This unified decoder preserves pre-trained vision-language priors, supports parallel decoding, breaks the autoregressive bottleneck, and reduces the number of function evaluations. Discrete Diffusion VLA achieves 96.3% avg. success rates on LIBERO, 71.2% visual matching on SimplerEnv-Fractal and 54.2% overall on SimplerEnv-Bridge. We also provide ablation study on vision-language ability retention on LIBERO-OOD (Out-of-Distribution) benchmark, with our method improving over autoregressive, MLP decoder and continuous diffusion baselines. These findings indicate that discrete-diffusion VLA supports precise action modeling and consistent training, laying groundwork for scaling VLA to larger models and datasets. Our code is available at https://github.com/Liang-ZX/DiscreteDiffusionVLA/tree/libero.
URLs: https://github.com/Liang-ZX/DiscreteDiffusionVLA/tree/libero.
Authors: Yangsong Zhang, Abdul Ahad Butt, G\"ul Varol, Ivan Laptev
Abstract: Human motion generation has shown great advances thanks to the recent diffusion models trained on large-scale motion capture data. Most of existing works, however, currently target animation of isolated people in empty scenes. Meanwhile, synthesizing realistic human-object interactions in complex 3D scenes remains a critical challenge in computer graphics and robotics. One obstacle towards generating versatile high-fidelity human-object interactions is the lack of large-scale datasets with diverse object manipulations. Indeed, existing motion capture data is typically restricted to single people and manipulations of limited sets of objects. To address this issue, we propose an automatic motion extraction pipeline and use it to collect interaction-rich human motions. Our new dataset InterPose contains 73.8K sequences of 3D human motions and corresponding text captions automatically obtained from 45.8K videos with human-object interactions. We perform extensive experiments and demonstrate InterPose to bring significant improvements to state-of-the-art methods for human motion generation. Moreover, using InterPose we develop an LLM-based agent enabling zero-shot animation of people interacting with diverse objects and scenes.
Authors: Christian Geckeler, Niklas Neugebauer, Manasi Muglikar, Davide Scaramuzza, Stefano Mintchev
Abstract: Uncrewed aerial vehicles (UAVs) are increasingly deployed in forest environments for tasks such as environmental monitoring and search and rescue, which require safe navigation through dense foliage and precise data collection. Traditional sensing approaches, including passive multispectral and RGB imaging, suffer from latency, poor depth resolution, and strong dependence on ambient light - especially under forest canopies. In this work, we present a novel event spectroscopy system that simultaneously enables high-resolution, low-latency depth reconstruction with integrated multispectral imaging using a single sensor. Depth is reconstructed using structured light, and by modulating the wavelength of the projected structured light, our system captures spectral information in controlled bands between 650 nm and 850 nm. We demonstrate up to $60\%$ improvement in RMSE over commercial depth sensors and validate the spectral accuracy against a reference spectrometer and commercial multispectral cameras, demonstrating comparable performance. A portable version limited to RGB (3 wavelengths) is used to collect real-world depth and spectral data from a Masoala Rainforest. We demonstrate the use of this prototype for color image reconstruction and material differentiation between leaves and branches using spectral and depth data. Our results show that adding depth (available at no extra effort with our setup) to material differentiation improves the accuracy by over $30\%$ compared to color-only method. Our system, tested in both lab and real-world rainforest environments, shows strong performance in depth estimation, RGB reconstruction, and material differentiation - paving the way for lightweight, integrated, and robust UAV perception and data collection in complex natural environments.
Authors: Heng Zhang, Haichuan Hu, Yaomin Shen, Weihao Yu, Yilei Yuan, Haochen You, Guo Cheng, Zijian Zhang, Lubin Gan, Huihui Wei, Hao Zhang, Jin Huang
Abstract: Large Vision-Language Models (LVLMs) have demonstrated impressive performance on multimodal tasks through scaled architectures and extensive training. However, existing Mixture of Experts (MoE) approaches face challenges due to the asymmetry between visual and linguistic processing. Visual information is spatially complete, while language requires maintaining sequential context. As a result, MoE models struggle to balance modality-specific features and cross-modal interactions. Through systematic analysis, we observe that language experts in deeper layers progressively lose contextual grounding and rely more on parametric knowledge rather than utilizing the provided visual and linguistic information. To address this, we propose AsyMoE, a novel architecture that models this asymmetry using three specialized expert groups. We design intra-modality experts for modality-specific processing, hyperbolic inter-modality experts for hierarchical cross-modal interactions, and evidence-priority language experts to suppress parametric biases and maintain contextual grounding. Extensive experiments demonstrate that AsyMoE achieves 26.58% and 15.45% accuracy improvements over vanilla MoE and modality-specific MoE respectively, with 25.45% fewer activated parameters than dense models.
Authors: Alexandros Doumanoglou, Kurt Driessens, Dimitrios Zarpalas
Abstract: Empirical evidence shows that deep vision networks often represent concepts as directions in latent space with concept information written along directional components in the vector representation of the input. However, the mechanism to encode (write) and decode (read) concept information to and from vector representations is not directly accessible as it constitutes a latent mechanism that naturally emerges from the training process of the network. Recovering this mechanism unlocks significant potential to open the black-box nature of deep networks, enabling understanding, debugging, and improving deep learning models. In this work, we propose an unsupervised method to recover this mechanism. For each concept, we explain that under the hypothesis of linear concept representations, this mechanism can be implemented with the help of two directions: the first facilitating encoding of concept information and the second facilitating decoding. Unlike prior matrix decomposition, autoencoder, or dictionary learning methods that rely on feature reconstruction, we propose a new perspective: decoding directions are identified via directional clustering of activations, and encoding directions are estimated with signal vectors under a probabilistic view. We further leverage network weights through a novel technique, Uncertainty Region Alignment, which reveals interpretable directions affecting predictions. Our analysis shows that (a) on synthetic data, our method recovers ground-truth direction pairs; (b) on real data, decoding directions map to monosemantic, interpretable concepts and outperform unsupervised baselines; and (c) signal vectors faithfully estimate encoding directions, validated via activation maximization. Finally, we demonstrate applications in understanding global model behavior, explaining individual predictions, and intervening to produce counterfactuals or correct errors.
Authors: Ioana Ciuclea, Giorgio Longari, Alice Barbara Tumpach
Abstract: Most datasets encountered in computer vision and medical applications present symmetries that should be taken into account in classification tasks. A typical example is the symmetry by rotation and/or scaling in object detection. A common way to build neural networks that learn the symmetries is to use data augmentation. In order to avoid data augmentation and build more sustainable algorithms, we present an alternative method to mod out symmetries based on the notion of section of a principal fiber bundle. This framework allows the use of simple metrics on the space of objects in order to measure dissimilarities between orbits of objects under the symmetry group. Moreover, the section used can be optimized to maximize separation of classes. We illustrate this methodology on a dataset of contours of objects for the groups of translations, rotations, scalings and reparameterizations. In particular, we present a $2$-parameter family of canonical parameterizations of curves, containing the constant-speed parameterization as a special case, which we believe is interesting in its own right. We hope that this simple application will serve to convey the geometric concepts underlying this method, which have a wide range of possible applications. The code is available at the following link: $\href{https://github.com/GiLonga/Geometric-Learning}{https://github.com/GiLonga/Geometric-Learning}$. A tutorial notebook showcasing an application of the code to a specific dataset is available at the following link: $\href{https://github.com/ioanaciuclea/geometric-learning-notebook}{https://github.com/ioanaciuclea/geometric-learning-notebook}$
URLs: https://github.com/GiLonga/Geometric-Learning, https://github.com/GiLonga/Geometric-Learning, https://github.com/ioanaciuclea/geometric-learning-notebook, https://github.com/ioanaciuclea/geometric-learning-notebook
Authors: Viktor Koz\'ak, Jan Chudoba, Libor P\v{r}eu\v{c}il
Abstract: An accurate and up-to-date model of a photovoltaic (PV) power plant is essential for its optimal operation and maintenance. However, such a model may not be easily available. This work introduces a novel approach for PV power plant mapping based on aerial overview images. It enables the automation of the mapping process while removing the reliance on third-party data. The presented mapping method takes advantage of the structural layout of the power plants to achieve detailed modeling down to the level of individual PV modules. The approach relies on visual segmentation of PV modules in overview images and the inference of structural information in each image, assigning modules to individual benches, rows, and columns. We identify visual keypoints related to the layout and use these to merge detections from multiple images while maintaining their structural integrity. The presented method was experimentally verified and evaluated on two different power plants. The final fusion of 3D positions and semantic structures results in a compact georeferenced model suitable for power plant maintenance.
Authors: Zhiwei Jin, Xiaohui Song, Nan Wang, Yafei Liu, Chao Li, Xin Li, Ruichen Wang, Zhihao Li, Qi Qi, Long Cheng, Dongze Hao, Quanlong Zheng, Yanhao Zhang, Haobo Ji, Jian Ma, Zhitong Zheng, Zhenyi Lin, Haolin Deng, Xin Zou, Xiaojie Yin, Ruilin Wang, Liankai Cai, Haijing Liu, Yuqing Qiu, Ke Chen, Zixian Li, Chi Xie, Huafei Li, Chenxing Li, Chuangchuang Wang, Kai Tang, Zhiguang Zhu, Kai Tang, Wenmei Gao, Rui Wang, Jun Wu, Chao Liu, Qin Xie, Chen Chen, Haonan Lu
Abstract: In recent years, while cloud-based MLLMs such as QwenVL, InternVL, GPT-4o, Gemini, and Claude Sonnet have demonstrated outstanding performance with enormous model sizes reaching hundreds of billions of parameters, they significantly surpass the limitations in memory, power consumption, and computing capacity of edge devices such as mobile phones. This paper introduces AndesVL, a suite of mobile-side MLLMs with 0.6B to 4B parameters based on Qwen3's LLM and various visual encoders. We comprehensively outline the model architectures, training pipeline, and training data of AndesVL, which achieves first-tier performance across a wide range of open-source benchmarks, including fields such as text-rich image understanding, reasoning and math, multi-image comprehension, general VQA, hallucination mitigation, multilingual understanding, and GUI-related tasks when compared with state-of-the-art models of a similar scale. Furthermore, we introduce a 1+N LoRA architecture alongside a Quantization-Aware LoRA Fine-Tuning (QALFT) framework to facilitate efficient task adaptation and model compression during mobile-side deployment of AndesVL. Moreover, utilizing our cache eviction algorithm -- OKV -- along with customized speculative decoding and compression strategies, we achieve a 6.7x peak decoding speedup ratio, up to 30.9% memory reduction, and 1.8 bits-per-weight when deploying AndesVL-4B on MediaTek Dimensity 9500 chips. We release all models on https://huggingface.co/OPPOer.
Authors: Yixuan Li, Yuhui Chen, Mingcai Zhou, Haoran Li, Zhengtao Zhang, Dongbin Zhao
Abstract: Spatial perception and reasoning are crucial for Vision-Language-Action (VLA) models to accomplish fine-grained manipulation tasks. However, existing approaches often lack the ability to understand and reason over the essential 3D structures necessary for precise control. To address this limitation, we propose QDepth-VLA, a general framework that augments VLA models with an auxiliary depth prediction task. A dedicated depth expert is designed to predict quantized latent tokens of depth maps obtained from a VQ-VAE encoder, enabling the model to learn depth-aware representations that capture critical geometric cues. Experimental results on the simulation benchmarks and real-world tasks demonstrate that QDepth-VLA yields strong spatial reasoning and competitive performance on manipulation tasks.
Authors: Xiaojun Guo, Runyu Zhou, Yifei Wang, Qi Zhang, Chenheng Zhang, Stefanie Jegelka, Xiaohan Wang, Jiajun Chai, Guojun Yin, Wei Lin, Yisen Wang
Abstract: Vision-language models (VLMs) have shown remarkable abilities by integrating large language models with visual inputs. However, they often fail to utilize visual evidence adequately, either depending on linguistic priors in vision-centric tasks or resorting to textual shortcuts during reasoning. Although reinforcement learning (RL) can align models with desired behaviors, its application to VLMs has been hindered by the lack of scalable and reliable reward mechanisms. To overcome this challenge, we propose SSL4RL, a novel framework that leverages self-supervised learning (SSL) tasks as a source of verifiable rewards for RL-based fine-tuning. Our approach reformulates SSL objectives-such as predicting image rotation or reconstructing masked patches-into dense, automatic reward signals, eliminating the need for human preference data or unreliable AI evaluators. Experiments show that SSL4RL substantially improves performance on both vision-centric and vision-language reasoning benchmarks. Furthermore, through systematic ablations, we identify key factors-such as task difficulty, model scale, and semantic alignment with the target domain-that influence the effectiveness of SSL4RL tasks, offering new design principles for future work. We also demonstrate the framework's generality by applying it to graph learning, where it yields significant gains. SSL4RL establishes a versatile and effective paradigm for aligning multimodal models using verifiable, self-supervised objectives.
Authors: Penghao Wang, Yiyang He, Xin Lv, Yukai Zhou, Lan Xu, Jingyi Yu, Jiayuan Gu
Abstract: Understanding objects at the level of their constituent parts is fundamental to advancing computer vision, graphics, and robotics. While datasets like PartNet have driven progress in 3D part understanding, their reliance on untextured geometries and expert-dependent annotation limits scalability and usability. We introduce PartNeXt, a next-generation dataset addressing these gaps with over 23,000 high-quality, textured 3D models annotated with fine-grained, hierarchical part labels across 50 categories. We benchmark PartNeXt on two tasks: (1) class-agnostic part segmentation, where state-of-the-art methods (e.g., PartField, SAMPart3D) struggle with fine-grained and leaf-level parts, and (2) 3D part-centric question answering, a new benchmark for 3D-LLMs that reveals significant gaps in open-vocabulary part grounding. Additionally, training Point-SAM on PartNeXt yields substantial gains over PartNet, underscoring the dataset's superior quality and diversity. By combining scalable annotation, texture-aware labels, and multi-task evaluation, PartNeXt opens new avenues for research in structured 3D understanding.
Authors: Xudong Yan, Songhe Feng
Abstract: Compositional Zero-Shot Learning (CZSL) aims to recognize novel attribute-object compositions based on the knowledge learned from seen ones. Existing methods suffer from performance degradation caused by the distribution shift of label space at test time, which stems from the inclusion of unseen compositions recombined from attributes and objects. To overcome the challenge, we propose a novel approach that accumulates comprehensive knowledge in both textual and visual modalities from unsupervised data to update multimodal prototypes at test time. Building on this, we further design an adaptive update weight to control the degree of prototype adjustment, enabling the model to flexibly adapt to distribution shift during testing. Moreover, a dynamic priority queue is introduced that stores high-confidence images to acquire visual knowledge from historical images for inference. Considering the semantic consistency of multimodal knowledge, we align textual and visual prototypes by multimodal collaborative representation learning. Extensive experiments indicate that our approach achieves state-of-the-art performance on four benchmark datasets under both closed-world and open-world settings. Code will be available at https://github.com/xud-yan/TOMCAT .
Authors: Wan Jiang, Jing Yan, Xiaojing Chen, Lin Shen, Chenhao Lin, Yunfeng Diao, Richang Hong
Abstract: Diffusion models have recently achieved remarkable photorealism, making it increasingly difficult to distinguish real images from generated ones, raising significant privacy and security concerns. In response, we present a key finding: structural edits enhance the reconstruction of real images while degrading that of generated images, creating a distinctive edit-induced reconstruction error shift. This asymmetric shift enhances the separability between real and generated images. Building on this insight, we propose EIRES, a training-free method that leverages structural edits to reveal inherent differences between real and generated images. To explain the discriminative power of this shift, we derive the reconstruction error lower bound under edit perturbations. Since EIRES requires no training, thresholding depends solely on the natural separability of the signal, where a larger margin yields more reliable detection. Extensive experiments show that EIRES is effective across diverse generative models and remains robust on the unbiased subset, even under post-processing operations.
Authors: Yinglu Li, Zhiying Lu, Zhihang Liu, Yiwei Sun, 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: Heng Zheng, Yuling Shi, Xiaodong Gu, Haochen You, Zijian Zhang, Lubin Gan, Hao Zhang, Wenjun Huang, Jin Huang
Abstract: Visual geo-localization requires extensive geographic knowledge and sophisticated reasoning to determine image locations without GPS metadata. Traditional retrieval methods are constrained by database coverage and quality. Recent Large Vision-Language Models (LVLMs) enable direct location reasoning from image content, yet individual models struggle with diverse geographic regions and complex scenes. Existing multi-agent systems improve performance through model collaboration but treat all agent interactions uniformly. They lack mechanisms to handle conflicting predictions effectively. We propose \textbf{GraphGeo}, a multi-agent debate framework using heterogeneous graph neural networks for visual geo-localization. Our approach models diverse debate relationships through typed edges, distinguishing supportive collaboration, competitive argumentation, and knowledge transfer. We introduce a dual-level debate mechanism combining node-level refinement and edge-level argumentation modeling. A cross-level topology refinement strategy enables co-evolution between graph structure and agent representations. Experiments on multiple benchmarks demonstrate GraphGeo significantly outperforms state-of-the-art methods. Our framework transforms cognitive conflicts between agents into enhanced geo-localization accuracy through structured debate.
Authors: Seyed Alireza Javid, Amirhossein Bagheri, Nuria Gonz\'alez-Prelcic
Abstract: Classifier-guided diffusion models have emerged as a powerful approach for conditional image generation, but they suffer from overconfident predictions during early denoising steps, causing the guidance gradient to vanish. This paper introduces two complementary contributions to address this issue. First, we propose a differentiable calibration objective based on the Smooth Expected Calibration Error (Smooth ECE), which improves classifier calibration with minimal fine-tuning and yields measurable improvements in Frechet Inception Distance (FID). Second, we develop enhanced sampling guidance methods that operate on off-the-shelf classifiers without requiring retraining. These include tilted sampling with batch-level reweighting, adaptive entropy-regularized sampling to preserve diversity, and a novel f-divergence-based sampling strategy that strengthens class-consistent guidance while maintaining mode coverage. Experiments on ImageNet 128x128 demonstrate that our divergence-regularized guidance achieves an FID of 2.13 using a ResNet-101 classifier, improving upon existing classifier-guided diffusion methods while requiring no diffusion model retraining. The results show that principled calibration and divergence-aware sampling provide practical and effective improvements for classifier-guided diffusion.
Authors: Alomar Antonia, Rubio Ricardo, Albaiges Gerard, Salort-Benejam Laura, Caminal Julia, Prat Maria, Rueda Carolina, Cortes Berta, Piella Gemma, Sukno Federico
Abstract: The automatic localization and standardization of anatomical planes in 3D medical imaging remains a challenging problem due to variability in object pose, appearance, and image quality. In 3D ultrasound, these challenges are exacerbated by speckle noise and limited contrast, particularly in fetal imaging. To address these challenges in the context of facial assessment, we present: 1) GT++, a robust algorithm that estimates standard facial planes from 3D US volumes using annotated anatomical landmarks; and 2) 3DFETUS, a deep learning model that automates and standardizes their localization in 3D fetal US volumes. We evaluated our methods both qualitatively, through expert clinical review, and quantitatively. The proposed approach achieved a mean translation error of 3.21 $\pm$ 1.98mm and a mean rotation error of 5.31 $\pm$ 3.945$^\circ$ per plane, outperforming other state-of-the-art methods on 3D US volumes. Clinical assessments further confirmed the effectiveness of both GT++ and 3DFETUS, demonstrating statistically significant improvements in plane estimation accuracy.
Authors: Weihua Wang, Yubo Cui, Xiangru Lin, Zhiheng Li, Zheng Fang
Abstract: Vision-based 3D Semantic Scene Completion (SSC) has received growing attention due to its potential in autonomous driving. While most existing approaches follow an ego-centric paradigm by aggregating and diffusing features over the entire scene, they often overlook fine-grained object-level details, leading to semantic and geometric ambiguities, especially in complex environments. To address this limitation, we propose Ocean, an object-centric prediction framework that decomposes the scene into individual object instances to enable more accurate semantic occupancy prediction. Specifically, we first employ a lightweight segmentation model, MobileSAM, to extract instance masks from the input image. Then, we introduce a 3D Semantic Group Attention module that leverages linear attention to aggregate object-centric features in 3D space. To handle segmentation errors and missing instances, we further design a Global Similarity-Guided Attention module that leverages segmentation features for global interaction. Finally, we propose an Instance-aware Local Diffusion module that improves instance features through a generative process and subsequently refines the scene representation in the BEV space. Extensive experiments on the SemanticKITTI and SSCBench-KITTI360 benchmarks demonstrate that Ocean achieves state-of-the-art performance, with mIoU scores of 17.40 and 20.28, respectively.
Authors: Harold Haodong Chen, Disen Lan, Wen-Jie Shu, Qingyang Liu, Zihan Wang, Sirui Chen, Wenkai Cheng, Kanghao Chen, Hongfei Zhang, Zixin Zhang, Rongjin Guo, Yu Cheng, Ying-Cong Chen
Abstract: The rapid evolution of video generative models has shifted their focus from producing visually plausible outputs to tackling tasks requiring physical plausibility and logical consistency. However, despite recent breakthroughs such as Veo 3's chain-of-frames reasoning, it remains unclear whether these models can exhibit reasoning capabilities similar to large language models (LLMs). Existing benchmarks predominantly evaluate visual fidelity and temporal coherence, failing to capture higher-order reasoning abilities. To bridge this gap, we propose TiViBench, a hierarchical benchmark specifically designed to evaluate the reasoning capabilities of image-to-video (I2V) generation models. TiViBench systematically assesses reasoning across four dimensions: i) Structural Reasoning & Search, ii) Spatial & Visual Pattern Reasoning, iii) Symbolic & Logical Reasoning, and iv) Action Planning & Task Execution, spanning 24 diverse task scenarios across 3 difficulty levels. Through extensive evaluations, we show that commercial models (e.g., Sora 2, Veo 3.1) demonstrate stronger reasoning potential, while open-source models reveal untapped potential that remains hindered by limited training scale and data diversity. To further unlock this potential, we introduce VideoTPO, a simple yet effective test-time strategy inspired by preference optimization. By performing LLM self-analysis on generated candidates to identify strengths and weaknesses, VideoTPO significantly enhances reasoning performance without requiring additional training, data, or reward models. Together, TiViBench and VideoTPO pave the way for evaluating and advancing reasoning in video generation models, setting a foundation for future research in this emerging field.
Authors: Patrick Amadeus Irawan, Ikhlasul Akmal Hanif, Muhammad Dehan Al Kautsar, Genta Indra Winata, Fajri Koto, Alham Fikri Aji
Abstract: Although the cultural dimension has been one of the key aspects in evaluating Vision-Language Models (VLMs), their ability to remain stable across diverse cultural inputs remains largely untested, despite being crucial to support diversity and multicultural societies. Existing evaluations often rely on benchmarks featuring only a singular cultural concept per image, overlooking scenarios where multiple, potentially unrelated cultural cues coexist. To address this gap, we introduce ConfusedTourist, a novel cultural adversarial robustness suite designed to assess VLMs' stability against perturbed geographical cues. Our experiments reveal a critical vulnerability, where accuracy drops heavily under simple image-stacking perturbations and even worsens with its image-generation-based variant. Interpretability analyses further show that these failures stem from systematic attention shifts toward distracting cues, diverting the model from its intended focus. These findings highlight a critical challenge: visual cultural concept mixing can substantially impair even state-of-the-art VLMs, underscoring the urgent need for more culturally robust multimodal understanding.
Authors: Dohun Lim, Minji Kim, Jaewoon Lim, Sungchan Kim
Abstract: We propose BRIC, a novel test-time adaptation (TTA) framework that enables long-term human motion generation by resolving execution discrepancies between diffusion-based kinematic motion planners and reinforcement learning-based physics controllers. While diffusion models can generate diverse and expressive motions conditioned on text and scene context, they often produce physically implausible outputs, leading to execution drift during simulation. To address this, BRIC dynamically adapts the physics controller to noisy motion plans at test time, while preserving pre-trained skills via a loss function that mitigates catastrophic forgetting. In addition, BRIC introduces a lightweight test-time guidance mechanism that steers the diffusion model in the signal space without updating its parameters. By combining both adaptation strategies, BRIC ensures consistent and physically plausible long-term executions across diverse environments in an effective and efficient manner. We validate the effectiveness of BRIC on a variety of long-term tasks, including motion composition, obstacle avoidance, and human-scene interaction, achieving state-of-the-art performance across all tasks.
Authors: Davide Nadalini, Manuele Rusci, Elia Cereda, Luca Benini, Francesco Conti, Daniele Palossi
Abstract: Monocular depth estimation (MDE) plays a crucial role in enabling spatially-aware applications in Ultra-low-power (ULP) Internet-of-Things (IoT) platforms. However, the limited number of parameters of Deep Neural Networks for the MDE task, designed for IoT nodes, results in severe accuracy drops when the sensor data observed in the field shifts significantly from the training dataset. To address this domain shift problem, we present a multi-modal On-Device Learning (ODL) technique, deployed on an IoT device integrating a Greenwaves GAP9 MicroController Unit (MCU), a 80 mW monocular camera and a 8 x 8 pixel depth sensor, consuming $\approx$300mW. In its normal operation, this setup feeds a tiny 107 k-parameter $\mu$PyD-Net model with monocular images for inference. The depth sensor, usually deactivated to minimize energy consumption, is only activated alongside the camera to collect pseudo-labels when the system is placed in a new environment. Then, the fine-tuning task is performed entirely on the MCU, using the new data. To optimize our backpropagation-based on-device training, we introduce a novel memory-driven sparse update scheme, which minimizes the fine-tuning memory to 1.2 MB, 2.2x less than a full update, while preserving accuracy (i.e., only 2% and 1.5% drops on the KITTI and NYUv2 datasets). Our in-field tests demonstrate, for the first time, that ODL for MDE can be performed in 17.8 minutes on the IoT node, reducing the root mean squared error from 4.9 to 0.6m with only 3 k self-labeled samples, collected in a real-life deployment scenario.
Authors: Xiuli Bi, Die Xiao, Junchao Fan, Bin Xiao
Abstract: In recent years, Contrastive Language-Image Pretraining (CLIP) has been widely applied to Weakly Supervised Semantic Segmentation (WSSS) tasks due to its powerful cross-modal semantic understanding capabilities. This paper proposes a novel Semantic and Spatial Rectification (SSR) method to address the limitations of existing CLIP-based weakly supervised semantic segmentation approaches: over-activation in non-target foreground regions and background areas. Specifically, at the semantic level, the Cross-Modal Prototype Alignment (CMPA) establishes a contrastive learning mechanism to enforce feature space alignment across modalities, reducing inter-class overlap while enhancing semantic correlations, to rectify over-activation in non-target foreground regions effectively; at the spatial level, the Superpixel-Guided Correction (SGC) leverages superpixel-based spatial priors to precisely filter out interference from non-target regions during affinity propagation, significantly rectifying background over-activation. Extensive experiments on the PASCAL VOC and MS COCO datasets demonstrate that our method outperforms all single-stage approaches, as well as more complex multi-stage approaches, achieving mIoU scores of 79.5% and 50.6%, respectively.
Authors: Tang Haonan, Chen Yanjun, Jiang Lezhi
Abstract: The TrackNet series has established a strong baseline for fast-moving small object tracking in sports. However, existing iterations face significant limitations: V1-V3 struggle with occlusions due to a reliance on purely visual cues, while TrackNetV4, despite introducing motion inputs, suffers from directional ambiguity as its absolute difference method discards motion polarity. To overcome these bottlenecks, we propose TrackNetV5, a robust architecture integrating two novel mechanisms. First, to recover lost directional priors, we introduce the Motion Direction Decoupling (MDD) module. Unlike V4, MDD decomposes temporal dynamics into signed polarity fields, explicitly encoding both movement occurrence and trajectory direction. Second, we propose the Residual-Driven Spatio-Temporal Refinement (R-STR) head. Operating on a coarse-to-fine paradigm, this Transformer-based module leverages factorized spatio-temporal contexts to estimate a corrective residual, effectively recovering occluded targets. Extensive experiments on the TrackNetV2 dataset demonstrate that TrackNetV5 achieves a new state-of-the-art F1-score of 0.9859 and an accuracy of 0.9733, significantly outperforming previous versions. Notably, this performance leap is achieved with a marginal 3.7% increase in FLOPs compared to V4, maintaining real-time inference capabilities while delivering superior tracking precision.
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: Mai Tsujimoto, Junjue Wang, Weihao Xuan, Naoto Yokoya
Abstract: Three-dimensional geospatial analysis is critical for applications in urban planning, climate adaptation, and environmental assessment. However, current methodologies depend on costly, specialized sensors, such as LiDAR and multispectral sensors, which restrict global accessibility. Additionally, existing sensor-based and rule-driven methods struggle with tasks requiring the integration of multiple 3D cues, handling diverse queries, and providing interpretable reasoning. We present Geo3DVQA, a comprehensive benchmark that evaluates vision-language models (VLMs) in height-aware 3D geospatial reasoning from RGB imagery alone. Unlike conventional sensor-based frameworks, Geo3DVQA emphasizes realistic scenarios integrating elevation, sky view factors, and land cover patterns. The benchmark comprises 110k curated question-answer pairs across 16 task categories, including single-feature inference, multi-feature reasoning, and application-level analysis. Through a systematic evaluation of ten state-of-the-art VLMs, we reveal fundamental limitations in RGB-to-3D spatial reasoning. Our results further show that domain-specific instruction tuning consistently enhances model performance across all task categories, including height-aware and open-ended, application-oriented reasoning. Geo3DVQA provides a unified, interpretable framework for evaluating RGB-based 3D geospatial reasoning and identifies key challenges and opportunities for scalable 3D spatial analysis. The code and data are available at https://github.com/mm1129/Geo3DVQA.
Authors: Yuanye Liu, Hanxiao Zhang, Jiyao Liu, Nannan Shi, Yuxin Shi, Arif Mahmood, Murtaza Taj, Xiahai Zhuang
Abstract: Liver fibrosis represents a significant global health burden, necessitating accurate staging for effective clinical management. This report introduces the LiQA (Liver Fibrosis Quantification and Analysis) dataset, established as part of the CARE 2024 challenge. Comprising $440$ patients with multi-phase, multi-center MRI scans, the dataset is curated to benchmark algorithms for Liver Segmentation (LiSeg) and Liver Fibrosis Staging (LiFS) under complex real-world conditions, including domain shifts, missing modalities, and spatial misalignment. We further describe the challenge's top-performing methodology, which integrates a semi-supervised learning framework with external data for robust segmentation, and utilizes a multi-view consensus approach with Class Activation Map (CAM)-based regularization for staging. Evaluation of this baseline demonstrates that leveraging multi-source data and anatomical constraints significantly enhances model robustness in clinical settings.
Authors: Ryan Banks, Camila Lindoni Azevedo, Hongying Tang, Yunpeng Li
Abstract: Accurate understanding of anatomical structures is essential for reliably staging certain dental diseases. A way of introducing this within semantic segmentation models is by utilising hierarchy-aware methodologies. However, existing hierarchy-aware segmentation methods largely encode anatomical structure through the loss functions, providing weak and indirect supervision. We introduce a general framework that embeds an explicit anatomical hierarchy into semantic segmentation by coupling a recurrent, level-wise prediction scheme with restrictive output heads and top-down feature conditioning. At each depth of the class tree, the backbone is re-run on the original image concatenated with logits from the previous level. Child class features are conditioned using Feature-wise Linear Modulation of their parent class probabilities, to modulate child feature spaces for fine grained detection. A probabilistic composition rule enforces consistency between parent and descendant classes. Hierarchical loss combines per-level class weighted Dice and cross entropy loss and a consistency term loss, ensuring parent predictions are the sum of their children. We validate our approach on our proposed dataset, TL-pano, containing 194 panoramic radiographs with dense instance and semantic segmentation annotations, of tooth layers and alveolar bone. Utilising UNet and HRNet as donor models across a 5-fold cross validation scheme, the hierarchical variants consistently increase IoU, Dice, and recall, particularly for fine-grained anatomies, and produce more anatomically coherent masks. However, hierarchical variants also demonstrated increased recall over precision, implying increased false positives. The results demonstrate that explicit hierarchical structuring improves both performance and clinical plausibility, especially in low data dental imaging regimes.
Authors: Youming Deng, Songyou Peng, Junyi Zhang, Kathryn Heal, Tiancheng Sun, John Flynn, Steve Marschner, Lucy Chai
Abstract: Novel View Synthesis (NVS) has traditionally relied on models with explicit 3D inductive biases combined with known camera parameters from Structure-from-Motion (SfM) beforehand. Recent vision foundation models like VGGT take an orthogonal approach -- 3D knowledge is gained implicitly through training data and loss objectives, enabling feed-forward prediction of both camera parameters and 3D representations directly from a set of uncalibrated images. While flexible, VGGT features lack explicit multi-view geometric consistency, and we find that improving such 3D feature consistency benefits both NVS and pose estimation tasks. We introduce Selfi, a self-improving 3D reconstruction pipeline via feature alignment, transforming a VGGT backbone into a high-fidelity 3D reconstruction engine by leveraging its own outputs as pseudo-ground-truth. Specifically, we train a lightweight feature adapter using a reprojection-based consistency loss, which distills VGGT outputs into a new geometrically-aligned feature space that captures spatial proximity in 3D. This enables state-of-the-art performance in both NVS and camera pose estimation, demonstrating that feature alignment is a highly beneficial step for downstream 3D reasoning.
Authors: Liming Kuang, Yordanka Velikova, Mahdi Saleh, Jan-Nico Zaech, Danda Pani Paudel, Benjamin Busam
Abstract: Object pose estimation is a fundamental task in computer vision and robotics, yet most methods require extensive, dataset-specific training. Concurrently, large-scale vision language models show remarkable zero-shot capabilities. In this work, we bridge these two worlds by introducing ConceptPose, a framework for object pose estimation that is both training-free and model-free. ConceptPose leverages a vision-language-model (VLM) to create open-vocabulary 3D concept maps, where each point is tagged with a concept vector derived from saliency maps. By establishing robust 3D-3D correspondences across concept maps, our approach allows precise estimation of 6DoF relative pose. Without any object or dataset-specific training, our approach achieves state-of-the-art results on common zero shot relative pose estimation benchmarks, significantly outperforming existing methods by over 62% in ADD(-S) score, including those that utilize extensive dataset-specific training.
Authors: Jiahao Liu, Senhao Cao
Abstract: Large-scale Vision-Language models have achieved remarkable results in various domains, such as image captioning and conditioned image generation. Nevertheless, these models still encounter difficulties in achieving human-like compositional generalization. In this study, we propose a new method called Independent Density Estimation (IDE) to tackle this challenge. IDE aims to learn the connection between individual words in a sentence and the corresponding features in an image, enabling compositional generalization. We build two models based on the philosophy of IDE. The first one utilizes fully disentangled visual representations as input, and the second leverages a Variational Auto-Encoder to obtain partially disentangled features from raw images. Additionally, we propose an entropy-based compositional inference method to combine predictions of each word in the sentence. Our models exhibit superior generalization to unseen compositions compared to current models when evaluated on various datasets.
Authors: Dereje Shenkut, Vijayakumar Bhagavatula
Abstract: Multi-agent collaborative perception (CP) is a promising paradigm for improving autonomous driving safety, particularly for vulnerable road users like pedestrians, via robust 3D perception. However, existing CP approaches often optimize for vehicle detection performance metrics, underperforming on smaller, safety-critical objects such as pedestrians, where detection failures can be catastrophic. Furthermore, previous CP methods rely on full feature exchange rather than communicating only salient features that help reduce false negatives. To this end, we present FocalComm, a novel collaborative perception framework that focuses on exchanging hard-instance-oriented features among connected collaborative agents. FocalComm consists of two key novel designs: (1) a learnable progressive hard instance mining (HIM) module to extract hard instance-oriented features per agent, and (2) a query-based feature-level (intermediate) fusion technique that dynamically weights these identified features during collaboration. We show that FocalComm outperforms state-of-the-art collaborative perception methods on two challenging real-world datasets (V2X-Real and DAIR-V2X) across both vehicle-centric and infrastructure-centric collaborative setups. FocalComm also shows a strong performance gain in pedestrian detection in V2X-Real.
Authors: Mengyu Li, Xingcheng Zhou, Guang Chen, Alois Knoll, Hu Cao
Abstract: In Intelligent Transportation Systems (ITS), multi-object tracking is primarily based on frame-based cameras. However, these cameras tend to perform poorly under dim lighting and high-speed motion conditions. Event cameras, characterized by low latency, high dynamic range and high temporal resolution, have considerable potential to mitigate these issues. Compared to frame-based vision, there are far fewer studies on event-based vision. To address this research gap, we introduce an initial pilot dataset tailored for event-based ITS, covering vehicle and pedestrian detection and tracking. We establish a tracking-by-detection benchmark with a specialized feature extractor based on this dataset, achieving excellent performance.
Authors: Zihan Wang, Jiashun Wang, Jeff Tan, Yiwen Zhao, Jessica Hodgins, Shubham Tulsiani, Deva Ramanan
Abstract: We introduce CRISP, a method that recovers simulatable human motion and scene geometry from monocular video. Prior work on joint human-scene reconstruction relies on data-driven priors and joint optimization with no physics in the loop, or recovers noisy geometry with artifacts that cause motion tracking policies with scene interactions to fail. In contrast, our key insight is to recover convex, clean, and simulation-ready geometry by fitting planar primitives to a point cloud reconstruction of the scene, via a simple clustering pipeline over depth, normals, and flow. To reconstruct scene geometry that might be occluded during interactions, we make use of human-scene contact modeling (e.g., we use human posture to reconstruct the occluded seat of a chair). Finally, we ensure that human and scene reconstructions are physically-plausible by using them to drive a humanoid controller via reinforcement learning. Our approach reduces motion tracking failure rates from 55.2\% to 6.9\% on human-centric video benchmarks (EMDB, PROX), while delivering a 43\% faster RL simulation throughput. We further validate it on in-the-wild videos including casually-captured videos, Internet videos, and even Sora-generated videos. This demonstrates CRISP's ability to generate physically-valid human motion and interaction environments at scale, greatly advancing real-to-sim applications for robotics and AR/VR.
Authors: Jiacheng Cui, Bingkui Tong, Xinyue Bi, Xiaohan Zhao, Jiacheng Liu, Zhiqiang Shen
Abstract: Soft labels generated by teacher models have become a dominant paradigm for knowledge transfer and recent large-scale dataset distillation such as SRe2L, RDED, LPLD, offering richer supervision than conventional hard labels. However, we observe that when only a limited number of crops per image are used, soft labels are prone to local semantic drift: a crop may visually resemble another class, causing its soft embedding to deviate from the ground-truth semantics of the original image. This mismatch between local visual content and global semantic meaning introduces systematic errors and distribution misalignment between training and testing. In this work, we revisit the overlooked role of hard labels and show that, when appropriately integrated, they provide a powerful content-agnostic anchor to calibrate semantic drift. We theoretically characterize the emergence of drift under few soft-label supervision and demonstrate that hybridizing soft and hard labels restores alignment between visual content and semantic supervision. Building on this insight, we propose a new training paradigm, Hard Label for Alleviating Local Semantic Drift (HALD), which leverages hard labels as intermediate corrective signals while retaining the fine-grained advantages of soft labels. Extensive experiments on dataset distillation and large-scale conventional classification benchmarks validate our approach, showing consistent improvements in generalization. On ImageNet-1K, we achieve 42.7% with only 285M storage for soft labels, outperforming prior state-of-the-art LPLD by 9.0%. Our findings re-establish the importance of hard labels as a complementary tool, and call for a rethinking of their role in soft-label-dominated training.
Authors: Zhanwei Li, Liang Li, Jiawan Zhang
Abstract: High-quality annotated data plays a crucial role in achieving accurate segmentation. However, such data for medical image segmentation are often scarce due to the time-consuming and labor-intensive nature of manual annotation. To address this challenge, we propose the Adaptive Frequency Domain Alignment Network (AFDAN)--a novel domain adaptation framework designed to align features in the frequency domain and alleviate data scarcity. AFDAN integrates three core components to enable robust cross-domain knowledge transfer: an Adversarial Domain Learning Module that transfers features from the source to the target domain; a Source-Target Frequency Fusion Module that blends frequency representations across domains; and a Spatial-Frequency Integration Module that combines both frequency and spatial features to further enhance segmentation accuracy across domains. Extensive experiments demonstrate the effectiveness of AFDAN: it achieves an Intersection over Union (IoU) of 90.9% for vitiligo segmentation in the newly constructed VITILIGO2025 dataset and a competitive IoU of 82.6% on the retinal vessel segmentation benchmark DRIVE, surpassing existing state-of-the-art approaches.
Authors: Qiushuo Cheng, Jingjing Liu, Catherine Morgan, Alan Whone, Majid Mirmehdi
Abstract: The self-supervised pretraining paradigm has achieved great success in learning 3D action representations for skeleton-based action recognition using contrastive learning. However, learning effective representations for skeleton-based temporal action localization remains challenging and underexplored. Unlike video-level {action} recognition, detecting action boundaries requires temporally sensitive features that capture subtle differences between adjacent frames where labels change. To this end, we formulate a snippet discrimination pretext task for self-supervised pretraining, which densely projects skeleton sequences into non-overlapping segments and promotes features that distinguish them across videos via contrastive learning. Additionally, we build on strong backbones of skeleton-based action recognition models by fusing intermediate features with a U-shaped module to enhance feature resolution for frame-level localization. Our approach consistently improves existing skeleton-based contrastive learning methods for action localization on BABEL across diverse subsets and evaluation protocols. We also achieve state-of-the-art transfer learning performance on PKUMMD with pretraining on NTU RGB+D and BABEL.
Authors: Chaoyang Wang, Kaituo Feng, Dongyang Chen, Zhongyu Wang, Zhixun Li, Sicheng Gao, Meng Meng, Xu Zhou, Manyuan Zhang, Yuzhang Shang, Xiangyu Yue
Abstract: Recent advances have shown that multimodal large language models (MLLMs) benefit from multimodal interleaved chain-of-thought (CoT) with vision tool interactions. However, existing open-source models often exhibit blind tool-use reasoning patterns, invoking vision tools even when they are unnecessary, which significantly increases inference overhead and degrades model performance. To this end, we propose AdaTooler-V, an MLLM that performs adaptive tool-use by determining whether a visual problem truly requires tools. First, we introduce AT-GRPO, a reinforcement learning algorithm that adaptively adjusts reward scales based on the Tool Benefit Score of each sample, encouraging the model to invoke tools only when they provide genuine improvements. Moreover, we construct two datasets to support training: AdaTooler-V-CoT-100k for SFT cold start and AdaTooler-V-300k for RL with verifiable rewards across single-image, multi-image, and video data. Experiments across twelve benchmarks demonstrate the strong reasoning capability of AdaTooler-V, outperforming existing methods in diverse visual reasoning tasks. Notably, AdaTooler-V-7B achieves an accuracy of 89.8\% on the high-resolution benchmark V*, surpassing the commercial proprietary model GPT-4o and Gemini 1.5 Pro. All code, models, and data are released.
Authors: Chiao-An Yang, Ryo Hachiuma, Sifei Liu, Subhashree Radhakrishnan, Raymond A. Yeh, Yu-Chiang Frank Wang, Min-Hung Chen
Abstract: Despite advances in Multimodal LLMs (MLLMs), their ability to reason over 3D structures and temporal dynamics remains limited, constrained by weak 4D perception and temporal understanding. Existing 3D and 4D Video Question Answering (VQA) benchmarks also emphasize static scenes and lack region-level prompting. We tackle these issues by introducing: (a) 4D-RGPT, a specialized MLLM designed to capture 4D representations from video inputs with enhanced temporal perception; (b) Perceptual 4D Distillation (P4D), a training framework that transfers 4D representations from a frozen expert model into 4D-RGPT for comprehensive 4D perception; and (c) R4D-Bench, a benchmark for depth-aware dynamic scenes with region-level prompting, built via a hybrid automated and human-verified pipeline. Our 4D-RGPT achieves notable improvements on both existing 4D VQA benchmarks and the proposed R4D-Bench benchmark.
Authors: Bjorna Qesaraku, Jan Steckel
Abstract: Maritime autonomous systems require robust predictive capabilities to anticipate vessel motion and environmental dynamics. While transformer architectures have revolutionized AIS-based trajectory prediction and demonstrated feasibility for sonar frame forecasting, their application to maritime radar frame prediction remains unexplored, creating a critical gap given radar's all-weather reliability for navigation. This survey systematically reviews predictive modeling approaches relevant to maritime radar, with emphasis on transformer architectures for spatiotemporal sequence forecasting, where existing representative methods are analyzed according to data type, architecture, and prediction horizon. Our review shows that, while the literature has demonstrated transformer-based frame prediction for sonar sensing, no prior work addresses transformer-based maritime radar frame prediction, thereby defining a clear research gap and motivating a concrete research direction for future work in this area.
Authors: Wuyi Liu, Le Jin, Junxian Yang, Yuanchao Yu, Zishuo Peng, Jinfeng Xu, Xianzhi Li, Jun Zhou
Abstract: Automated defect inspection of assembled Printed Circuit Board Assemblies (PCBA) is quite challenging due to the insufficient labeled data, micro-defects with just a few pixels in visually-complex and high-resolution images. To address these challenges, we present HiSIR-Net, a High resolution, Self-supervised Reconstruction framework for pixel-wise PCBA localization. Our design combines two lightweight modules that make this practical on real 4K-resolution boards: (i) a Selective Input-Reconstruction Gate (SIR-Gate) that lets the model decide where to trust reconstruction versus the original input, thereby reducing irrelevant reconstruction artifacts and false alarms; and (ii) a Region-level Optimized Patch Selection (ROPS) scheme with positional cues to select overlapping patch reconstructions coherently across arbitrary resolutions. Organically integrating these mechanisms yields clean, high-resolution anomaly maps with low false positive (FP) rate. To bridge the gap in high-resolution PCBA datasets, we further contribute a self-collected dataset named SIPCBA-500 of 500 images. We conduct extensive experiments on our SIPCBA-500 as well as public benchmarks, demonstrating the superior localization performance of our method while running at practical speed. Full code and dataset will be made available upon acceptance.
Authors: Jiaze Li, Jingyang Chen, Yuxun Qu, Shijie Xu, Zhenru Lin, Junyou Zhu, Boshen Xu, Wenhui Tan, Pei Fu, Jianzhong Ju, Zhenbo Luo, Jian Luan
Abstract: We open-source MiMo-VL-Miloco-7B and its quantized variant MiMo-VL-Miloco-7B-GGUF, a pair of home-centric vision-language models that achieve strong performance on both home-scenario understanding and general multimodal reasoning. Built on the MiMo-VL-7B backbone, MiMo-VL-Miloco-7B is specialized for smart-home environments, attaining leading F1 scores on gesture recognition and common home-scenario understanding, while also delivering consistent gains across video benchmarks such as Video-MME, Video-MMMU, and Charades-STA, as well as language understanding benchmarks including MMMU-Pro and MMLU-Pro. In our experiments, MiMo-VL-Miloco-7B outperforms strong closed-source and open-source baselines on home-scenario understanding and several multimodal reasoning benchmarks. To balance specialization and generality, we design a two-stage training pipeline that combines supervised fine-tuning with reinforcement learning based on Group Relative Policy Optimization, leveraging efficient multi-domain data. We further incorporate chain-of-thought supervision and token-budget-aware reasoning, enabling the model to learn knowledge in a data-efficient manner while also performing reasoning efficiently. Our analysis shows that targeted home-scenario training not only enhances activity and gesture understanding, but also improves text-only reasoning with only modest trade-offs on document-centric tasks. Model checkpoints, quantized GGUF weights, and our home-scenario evaluation toolkit are publicly available at https://github.com/XiaoMi/xiaomi-mimo-vl-miloco to support research and deployment in real-world smart-home applications.
Authors: Yue Li, Qi Ma, Runyi Yang, Mengjiao Ma, Bin Ren, Nikola Popovic, Nicu Sebe, Theo Gevers, Luc Van Gool, Danda Pani Paudel, Martin R. Oswald
Abstract: While 3DGS has emerged as a high-fidelity scene representation, encoding rich, general-purpose features directly from its primitives remains under-explored. We address this gap by introducing Chorus, a multi-teacher pretraining framework that learns a holistic feed-forward 3D Gaussian Splatting (3DGS) scene encoder by distilling complementary signals from 2D foundation models. Chorus employs a shared 3D encoder and teacher-specific projectors to learn from language-aligned, generalist, and object-aware teachers, encouraging a shared embedding space that captures signals from high-level semantics to fine-grained structure. We evaluate Chorus on a wide range of tasks: open-vocabulary semantic and instance segmentation, linear and decoder probing, as well as data-efficient supervision. Besides 3DGS, we also test Chorus on several benchmarks that only support point clouds by pretraining a variant using only Gaussians' centers, colors, estimated normals as inputs. Interestingly, this encoder shows strong transfer and outperforms the point clouds baseline while using 39.9 times fewer training scenes. Finally, we propose a render-and-distill adaptation that facilitates out-of-domain finetuning. Our code and model will be released upon publication.
Authors: Juan Song, Jiaxiang He, Lijie Yang, Mingtao Feng, Keyan Wang
Abstract: Diffusion probabilistic models have recently achieved remarkable success in generating high-quality images. However, balancing high perceptual quality and low distortion remains challenging in application of diffusion models in image compression. To address this issue, we propose a novel Uncertainty-Guided image compression approach with wavelet Diffusion (UGDiff). Our approach focuses on high frequency compression via the wavelet transform, since high frequency components are crucial for reconstructing image details. We introduce a wavelet conditional diffusion model for high frequency prediction, followed by a residual codec that compresses and transmits prediction residuals to the decoder. This diffusion prediction-then-residual compression paradigm effectively addresses the low fidelity issue common in direct reconstructions by existing diffusion models. Considering the uncertainty from the random sampling of the diffusion model, we further design an uncertainty-weighted rate-distortion (R-D) loss tailored for residual compression, providing a more rational trade-off between rate and distortion. Comprehensive experiments on two benchmark datasets validate the effectiveness of UGDiff, surpassing state-of-the-art image compression methods in R-D performance, perceptual quality, subjective quality, and inference time. Our code is available at: https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main.
URLs: https://github.com/hejiaxiang1/Wavelet-Diffusion/tree/main.
Authors: Danilo Costarelli, Michele Piconi
Abstract: In this paper, we provide two algorithms based on the theory of multidimensional neural network (NN) operators activated by hyperbolic tangent sigmoidal functions. Theoretical results are recalled to justify the performance of the here implemented algorithms. Specifically, the first algorithm models multidimensional signals (such as digital images), while the second one addresses the problem of rescaling and enhancement of the considered data. We discuss several applications of the NN-based algorithms for modeling and rescaling/enhancement remote sensing data (represented as images), with numerical experiments conducted on a selection of remote sensing (RS) images from the (open access) RETINA dataset. A comparison with classical interpolation methods, such as bilinear and bicubic interpolation, shows that the proposed algorithms outperform the others, particularly in terms of the Structural Similarity Index (SSIM).
Authors: Talon Chandler, Ivan E. Ivanov, Gabriel Sturm, Sheng Xiao, Xiang Zhao, Alexander Hillsley, Allyson Quinn Ryan, Ziwen Liu, Sricharan Reddy Varra, Ilan Theodoro, Eduardo Hirata-Miyasaki, Deepika Sundarraman, Amitabh Verma, Madhurya Sekhar, Chad Liu, Soorya Pradeep, See-Chi Lee, Shannon N. Rhoads, Maria Clara Zanellati, Sarah Cohen, Carolina Arias, Manuel D. Leonetti, Adrian Jacobo, Keir Balla, Lo\"ic A. Royer, Shalin B. Mehta
Abstract: Correlative computational microscopy can accelerate imaging and modeling of cellular dynamics by relaxing trade-offs inherent to dynamic imaging. Existing computational microscopy frameworks are either specialized or overly generic, limiting use to fixed configurations or domain experts. We introduce WaveOrder, a generalist wave-optical framework for imaging the architectural order of biomolecules. WaveOrder reconstructs diverse specimen properties from multi-channel acquisitions, with or without fluorescence. It provides a unified representation of linear optical properties and differentiable physics-based image formation models spanning widefield, confocal, light-sheet, and oblique label-free geometries. WaveOrder uses physics-informed ML to auto-tune model parameters and solve blind shift-variant restoration problems. This open-source, PyTorch-based framework enables scalable quantitative imaging across scales from organelles to adult zebrafish, and improves restoration of cellular structures in high-throughput experiments. We validate WaveOrder on diverse imaging applications, demonstrating its ability to recover biomolecular structure beyond the limits of existing approaches.
Authors: Weizhi Xian, Mingliang Zhou, Leong Hou U, Zhengguo Li
Abstract: In this paper, we propose a Physical Imaging Guided perceptual framework for Underwater Image Quality Assessment (UIQA), termed PIGUIQA. First, we formulate UIQA as a comprehensive problem that considers the combined effects of direct transmission attenuation and backward scattering on image perception. By leveraging underwater radiative transfer theory, we systematically integrate physics-based imaging estimations to establish quantitative metrics for these distortions. Second, recognizing spatial variations in image content significance and human perceptual sensitivity to distortions, we design a module built upon a neighborhood attention mechanism for local perception of images. This module effectively captures subtle features in images, thereby enhancing the adaptive perception of distortions on the basis of local information. Third, by employing a global perceptual aggregator that further integrates holistic image scene with underwater distortion information, the proposed model accurately predicts image quality scores. Extensive experiments across multiple benchmarks demonstrate that PIGUIQA achieves state-of-the-art performance while maintaining robust cross-dataset generalizability. The implementation is publicly available at https://github.com/WeizhiXian/PIGUIQA
Authors: Xiang Tang, Ruotong Li, Xiaopeng Fan
Abstract: In recent years, the demand for 3D content has grown exponentially with the intelligent upgrade of interactive media, extended reality (XR), and Metaverse industries. In order to overcome the limitations of traditional manual modeling approaches, such as labor-intensive workflows and prolonged production cycles, revolutionary advances have been achieved through the convergence of novel 3D representation paradigms and artificial intelligence generative technologies. In this survey, we conduct a systematic review of the cutting-edge achievements in static 3D object and scene generation, as well as establish a comprehensive technical framework through systematic categorization. We start our analysis with mainstream 3D object representations. Subsequently, we delve into the technical pathways of 3D object generation based on four mainstream deep generative models: Variational Autoencoders, Generative Adversarial Networks, Autoregressive Models, and Diffusion Models. Regarding scene generation, we focus on three dominant paradigms: layout-guided generation, lifting based on 2D priors, and rule-driven modeling. Finally, we critically examine persistent challenges in 3D generation and propose potential research directions for future investigation. This survey aims to provide readers with a structured understanding of state-of-the-art 3D generation technologies while inspiring researchers to undertake more exploration in this domain.
Authors: Daniel J. Strick, Carlos Garcia, Anthony Huang, Thomas Gardos
Abstract: Deep learning for radiologic image analysis is a rapidly growing field in biomedical research and is likely to become a standard practice in modern medicine. On the publicly available NIH ChestX-ray14 dataset, containing X-ray images that are classified by the presence or absence of 14 different diseases, we reproduced an algorithm known as CheXNet, as well as explored other algorithms that outperform CheXNet's baseline metrics. Model performance was primarily evaluated using the F1 score and AUC-ROC, both of which are critical metrics for imbalanced, multi-label classification tasks in medical imaging. The best model achieved an average AUC-ROC score of 0.85 and an average F1 score of 0.39 across all 14 disease classifications present in the dataset.
Authors: Bowen Feng, Zhiting Mei, Baiang Li, Julian Ost, Filippo Ghilotti, Roger Girgis, Anirudha Majumdar, Felix Heide
Abstract: While autonomous driving (AD) stacks struggle with decision making under partial observability and real-world complexity, human drivers are capable of commonsense reasoning to make near-optimal decisions with limited information. Recent work has attempted to leverage finetuned Vision-Language Models (VLMs) for trajectory planning at inference time to emulate human behavior. Despite their success in benchmark evaluations, these methods are often impractical to deploy (a 70B parameter VLM inference at merely 8 tokens per second requires more than 160G of memory), and their monolithic network structure prohibits safety decomposition. To bridge this gap, we propose VLM-Embedded Reasoning for autonomous Driving (VERDI), a training-time framework that distills the reasoning process and commonsense knowledge of VLMs into the AD stack. VERDI augments modular differentiable end-to-end (e2e) AD models by aligning intermediate module outputs at the perception, prediction, and planning stages with text features explaining the driving reasoning process produced by VLMs. By encouraging alignment in latent space, VERDI enables the modular AD stack to internalize structured reasoning, without incurring the inference-time costs of large VLMs. We validate VERDI in both open-loop (NuScenes and Bench2Drive benchmarks) and closed-loop (HugSim Simulator) settings. We find that VERDI outperforms existing e2e methods that do not embed reasoning by up to 11% in $\ell_{2}$ distance and 11% in driving performance, while maintaining real-time inference speed.
Authors: Kirill Muravyev, Artem Kobozev, Vasily Yuryev, Alexander Melekhin, Oleg Bulichev, Dmitry Yudin, Konstantin Yakovlev
Abstract: We propose PRISM-Loc - a lightweight and robust approach for localization in large outdoor environments that combines a compact topological representation with a novel scan-matching and curb-detection module operating on raw LiDAR scans. The method is designed for resource-constrained platforms and emphasizes real-time performance and resilience to common urban sensing challenges. It provides accurate localization in compact topological maps using global place recognition and an original scan matching technique. Experiments on standard benchmarks and on an embedded platform demonstrate the effectiveness of our approach. Our method achieves a 99\% success rate on the large-scale ITLP-Campus dataset while running at 150 ms per localization and using a 20 MB map for localization. We highlight three main contributions: (1) a compact representation for city-scale localization; (2) a novel curb detection and scan matching pipeline operating directly on raw LiDAR points; (3) a thorough evaluation of our method with performance analysis.
Authors: Ronald Skorobogat, Karsten Roth, Mariana-Iuliana Georgescu
Abstract: Model merging enables the combination of multiple specialized expert models into a single model capable of performing multiple tasks. However, the benefits of merging an increasing amount of specialized experts generally lead to diminishing returns and reduced overall performance gains. In this work, we empirically and theoretically analyze this limitation, proving that for Task Arithmetic-based methods, as more experts are merged, the common information dominates the task-specific information, leading to inevitable rank collapse. To mitigate this issue, we introduce Subspace Boosting, which operates on the singular value decomposed task vector space and maintains task vector ranks. Subspace Boosting raises merging efficacy for up to 20 experts by large margins of more than 10% when evaluated on both vision and language benchmarks. Moreover, we propose employing Higher-Order Generalized Singular Value Decomposition to quantify task similarity, offering a new interpretable perspective on model merging. Code and models are available at https://github.com/ronskoro/Subspace-Boosting.
Authors: Yuzhuo Chen, Zehua Ma, Han Fang, Weiming Zhang, Nenghai Yu
Abstract: AI-generated content (AIGC) enables efficient visual creation but raises copyright and authenticity risks. As a common technique for integrity verification and source tracing, digital image watermarking is regarded as a potential solution to above issues. However, the widespread adoption and advancing capabilities of generative image editing tools have amplified malicious tampering risks, while simultaneously posing new challenges to passive tampering detection and watermark robustness. To address these challenges, this paper proposes a Tamper-Aware Generative image WaterMarking method named TAG-WM. The proposed method comprises four key modules: a dual-mark joint sampling (DMJS) algorithm for embedding copyright and localization watermarks into the latent space while preserving generative quality, the watermark latent reconstruction (WLR) utilizing reversed DMJS, a dense variation region detector (DVRD) leveraging diffusion inversion sensitivity to identify tampered areas via statistical deviation analysis, and the tamper-aware decoding (TAD) guided by localization results. The experimental results demonstrate that TAG-WM achieves state-of-the-art performance in both tampering robustness and localization capability even under distortion, while preserving lossless generation quality and maintaining a watermark capacity of 256 bits. The code is available at: https://github.com/Suchenl/TAG-WM.
Authors: Yao Feng, Hengkai Tan, Xinyi Mao, Chendong Xiang, Guodong Liu, Shuhe Huang, Hang Su, Jun Zhu
Abstract: Scaling general-purpose manipulation to new robot embodiments remains challenging: each platform typically needs large, homogeneous demonstrations, and end-to-end pixel-to-action pipelines may degenerate under background and viewpoint shifts. Based on previous advances in video-based robot control, we present Vidar, consisting of an embodied video diffusion model as the generalizable prior and a masked inverse dynamics model (MIDM) as the adapter. We leverage a video diffusion model pre-trained at Internet scale, and further continuously pre-train it for the embodied domain using 750K multi-view trajectories collected from three real-world robot platforms. For this embodied pre-training, we introduce a unified observation space that jointly encodes robot, camera, task, and scene contexts. The MIDM module learns action-relevant pixel masks without dense labels, grounding the prior into the target embodiment's action space while suppressing distractors. With only 20 minutes of human demonstrations on an unseen robot (1% of typical data), Vidar outperforms state-of-the-art baselines and generalizes to unseen tasks, backgrounds, and camera layouts. Our results suggest a scalable recipe for "one prior, many embodiments": strong, inexpensive video priors together with minimal on-robot alignment.
Authors: Jinzhi Wang, Jiangbo Zhang, Chenzhan Yu, Zhigang Xiu, Duwei Dai, Ziyu xu, Ningyong Wu, Wenhong Zhao
Abstract: The growing volume of medical imaging data has increased the need for automated diagnostic tools, especially for musculoskeletal injuries like rib fractures, commonly detected via CT scans. Manual interpretation is time-consuming and error-prone. We propose OrthoInsight, a multi-modal deep learning framework for rib fracture diagnosis and report generation. It integrates a YOLOv9 model for fracture detection, a medical knowledge graph for retrieving clinical context, and a fine-tuned LLaVA language model for generating diagnostic reports. OrthoInsight combines visual features from CT images with expert textual data to deliver clinically useful outputs. Evaluated on 28,675 annotated CT images and expert reports, it achieves high performance across Diagnostic Accuracy, Content Completeness, Logical Coherence, and Clinical Guidance Value, with an average score of 4.28, outperforming models like GPT-4 and Claude-3. This study demonstrates the potential of multi-modal learning in transforming medical image analysis and providing effective support for radiologists.
Authors: Jin Ye, Jingran Wang, Fengchao Xiong, Jingzhou Chen, Yuntao Qian
Abstract: Hyperspectral images (HSIs) play a crucial role in remote sensing but are often degraded by complex noise patterns. Ensuring the physical property of the denoised HSIs is vital for robust HSI denoising, giving the rise of deep unfolding-based methods. However, these methods map the optimization of a physical model to a learnable network with a predefined depth, which lacks convergence guarantees. In contrast, Deep Equilibrium (DEQ) models treat the hidden layers of deep networks as the solution to a fixed-point problem and models them as infinite-depth networks, naturally consistent with the optimization. Under the framework of DEQ, we propose a Deep Equilibrium Convolutional Sparse Coding (DECSC) framework that unifies local spatial-spectral correlations, nonlocal spatial self-similarities, and global spatial consistency for robust HSI denoising. Within the convolutional sparse coding (CSC) framework, we enforce shared 2D convolutional sparse representation to ensure global spatial consistency across bands, while unshared 3D convolutional sparse representation captures local spatial-spectral details. To further exploit nonlocal self-similarities, a transformer block is embedded after the 2D CSC. Additionally, a detail enhancement module is integrated with the 3D CSC to promote image detail preservation. We formulate the proximal gradient descent of the CSC model as a fixed-point problem and transform the iterative updates into a learnable network architecture within the framework of DEQ. Experimental results demonstrate that our DECSC method achieves superior denoising performance compared to state-of-the-art methods.
Authors: Jan Nikolas Morshuis, Matthias Hein, Christian F. Baumgartner
Abstract: MR imaging is a valuable diagnostic tool allowing to non-invasively visualize patient anatomy and pathology with high soft-tissue contrast. However, MRI acquisition is typically time-consuming, leading to patient discomfort and increased costs to the healthcare system. Recent years have seen substantial research effort into the development of methods that allow for accelerated MRI acquisition while still obtaining a reconstruction that appears similar to the fully-sampled MR image. However, for many applications a perfectly reconstructed MR image may not be necessary, particularly, when the primary goal is a downstream task such as segmentation. This has led to growing interest in methods that aim to perform segmentation directly on accelerated MRI data. Despite recent advances, existing methods have largely been developed in isolation, without direct comparison to one another, often using separate or private datasets, and lacking unified evaluation standards. To date, no high-quality, comprehensive comparison of these methods exists, and the optimal strategy for segmenting accelerated MR data remains unknown. This paper provides the first unified benchmark for the segmentation of undersampled MRI data comparing 7 approaches. A particular focus is placed on comparing \textit{one-stage approaches}, that combine reconstruction and segmentation into a unified model, with \textit{two-stage approaches}, that utilize established MRI reconstruction methods followed by a segmentation network. We test these methods on two MRI datasets that include multi-coil k-space data as well as a human-annotated segmentation ground-truth. We find that simple two-stage methods that consider data-consistency lead to the best segmentation scores, surpassing complex specialized methods that are developed specifically for this task.
Authors: Luvin Munish Ragoo, Ivar Farup, Casper F. Andersen, Graham Finlayson
Abstract: This study investigates the impact of spectral filtering on color-matching functions (CMFs) and its implications for observer variability modeling. We conducted color matching experiments with two observers, both with and without a spectral filter in front of a bipartite field. Using a novel computational approach, we estimated the filter transmittance and transformation matrix necessary to convert unfiltered CMFs to filtered CMFs. Statistical analysis revealed good agreement between estimated and measured filter characteristics, particularly in central wavelength regions. Applying this methodology to compare between Stiles and Burch 1955 (SB1955) mean observer CMFs and our previously published "ICVIO" mean observer CMFs, we identified a "yellow" (short-wavelength suppressing) filter that effectively transforms between these datasets. This finding aligns with our hypothesis that observed differences between the CMF sets are attributable to age-related lens yellowing (average observer age: 49 years in ICVIO versus 30 years in SB1955). Our approach enables efficient representation of observer variability through a single filter rather than three separate functions, offering potentially reduced experimental overhead while maintaining accuracy in characterizing individual color vision differences.
Authors: Federico Stella, Nicolas Talabot, Hieu Le, Pascal Fua
Abstract: Unsigned Distance Fields (UDFs) are a natural implicit representation for open surfaces but, unlike Signed Distance Fields (SDFs), are challenging to triangulate into explicit meshes. This is especially true at high resolutions where neural UDFs exhibit higher noise levels, which makes it hard to capture fine details. Most current techniques perform within single voxels without reference to their neighborhood, resulting in missing surface and holes where the UDF is ambiguous or noisy. We show that this can be remedied by performing several passes and by reasoning on previously extracted surface elements to incorporate neighborhood information. Our key contribution is an iterative neural network that does this and progressively improves surface recovery within each voxel by spatially propagating information from increasingly distant neighbors. Unlike single-pass methods, our approach integrates newly detected surfaces, distance values, and gradients across multiple iterations, effectively correcting errors and stabilizing extraction in challenging regions. Experiments on diverse 3D models demonstrate that our method produces significantly more accurate and complete meshes than existing approaches, particularly for complex geometries, enabling UDF surface extraction at higher resolutions where traditional methods fail.
Authors: Kai Zhang, Corey D Barrett, Jangwon Kim, Lichao Sun, Tara Taghavi, Krishnaram Kenthapadi
Abstract: Agentic systems offer a potential path to solve complex clinical tasks through collaboration among specialized agents, augmented by tool use and external knowledge bases. Nevertheless, for chest X-ray (CXR) interpretation, prevailing methods remain limited: (i) reasoning is frequently neither clinically interpretable nor aligned with guidelines, reflecting mere aggregation of tool outputs; (ii) multimodal evidence is insufficiently fused, yielding text-only rationales that are not visually grounded; and (iii) systems rarely detect or resolve cross-tool inconsistencies and provide no principled verification mechanisms. To bridge the above gaps, we present RadAgents, a multi-agent framework that couples clinical priors with task-aware multimodal reasoning and encodes a radiologist-style workflow into a modular, auditable pipeline. In addition, we integrate grounding and multimodal retrieval-augmentation to verify and resolve context conflicts, resulting in outputs that are more reliable, transparent, and consistent with clinical practice.
Authors: Anusha Agarwal, Dibakar Roy Sarkar, Somdatta Goswami
Abstract: Background: Traumatic brain injury (TBI) is a major global health concern with 69 million annual cases. While neural operators have revolutionized scientific computing, existing architectures cannot handle the heterogeneous multimodal data (anatomical imaging, scalar demographics, and geometric constraints) required for patient-specific biomechanical modeling. Objective: This study introduces the first multimodal neural operator framework for biomechanics, fusing heterogeneous inputs to predict brain displacement fields for rapid TBI risk assessment. Methods: TBI modeling was reformulated as a multimodal operator learning problem. We proposed two fusion strategies: field projection for Fourier Neural Operator (FNO) architectures and branch decomposition for Deep Operator Networks (DeepONet). Four architectures (FNO, Factorized FNO, Multi-Grid FNO, and DeepONet) were extended with fusion mechanisms and evaluated on 249 in vivo Magnetic Resonance Elastography (MRE) datasets (20-90 Hz). Results: Multi-Grid FNO achieved the highest accuracy (MSE = 0.0023, 94.3% spatial fidelity). DeepONet offered the fastest inference (14.5 iterations/s, 7x speedup), suitable for edge deployment. All architectures reduced computation from hours to milliseconds. Conclusion: Multimodal neural operators enable efficient, real-time, patient-specific TBI risk assessment. This framework establishes a generalizable paradigm for heterogeneous data fusion in scientific domains, including precision medicine.
Authors: Danial Hosseintabar, Fan Chen, Giannis Daras, Antonio Torralba, Constantinos Daskalakis
Abstract: Diffusion models have emerged as powerful generative priors for high-dimensional inverse problems, yet learning them when only corrupted or noisy observations are available remains challenging. In this work, we propose a new method for training diffusion models with Expectation-Maximization (EM) from corrupted data. Our proposed method, DiffEM, utilizes conditional diffusion models to reconstruct clean data from observations in the E-step, and then uses the reconstructed data to refine the conditional diffusion model in the M-step. Theoretically, we provide monotonic convergence guarantees for the DiffEM iteration, assuming appropriate statistical conditions. We demonstrate the effectiveness of our approach through experiments on various image reconstruction tasks.
Authors: Yash Mittal, Dmitry Ignatov, Radu Timofte
Abstract: It introduces FractalNet, a fractal-inspired computational architectures for advanced large language model analysis that mainly challenges model diversity on a large scale in an efficient manner. The new set-up involves a template-driven generator, runner, and evaluation framework that, through systematic permutations of convolutional, normalization, activation, and dropout layers, can create more than 1,200 variants of neural networks. Fractal templates allow for structural recursion and multi-column pathways, thus, models become deeper and wider in a balanced way. Training utilizes PyTorch, Automatic Mixed Precision (AMP), and gradient checkpointing and is carried out on the CIFAR-10 dataset for five epochs. The outcomes show that fractal-based architectures are capable of strong performance and are computationally efficient. The paper positions fractal design as a feasible and resource-efficient method of automated architecture exploration.
Authors: Yiyun Zhou, Mingjing Xu, Jingwei Shi, Quanjiang Li, Jingyuan Chen
Abstract: Tactile sensing offers rich and complementary information to vision and language, enabling robots to perceive fine-grained object properties. However, existing tactile sensors lack standardization, leading to redundant features that hinder cross-sensor generalization. Moreover, existing methods fail to fully integrate the intermediate communication among tactile, language, and vision modalities. To address this, we propose TLV-CoRe, a CLIP-based Tactile-Language-Vision Collaborative Representation learning method. TLV-CoRe introduces a Sensor-Aware Modulator to unify tactile features across different sensors and employs tactile-irrelevant decoupled learning to disentangle irrelevant tactile features. Additionally, a Unified Bridging Adapter is introduced to enhance tri-modal interaction within the shared representation space. To fairly evaluate the effectiveness of tactile models, we further propose the RSS evaluation framework, focusing on Robustness, Synergy, and Stability across different methods. Experimental results demonstrate that TLV-CoRe significantly improves sensor-agnostic representation learning and cross-modal alignment, offering a new direction for multimodal tactile representation.
Authors: Xiuxiu Qi, Yu Yang, Jiannong Cao, Luyao Bai, Chongshan Fan, Chengtai Cao, Hongpeng Wang
Abstract: Language-conditioned manipulation facilitates human-robot interaction via behavioral cloning (BC), which learns control policies from human demonstrations and serves as a cornerstone of embodied AI. Overcoming compounding errors in sequential action decisions remains a central challenge to improving BC performance. Existing approaches mitigate compounding errors through data augmentation, expressive representation, or temporal abstraction. However, they suffer from physical discontinuities and semantic-physical misalignment, leading to inaccurate action cloning and intermittent execution. In this paper, we present Continuous vision-language-action Co-Learning with Semantic-Physical Alignment (CCoL), a novel BC framework that ensures temporally consistent execution and fine-grained semantic grounding. It generates robust and smooth action execution trajectories through continuous co-learning across vision, language, and proprioceptive inputs (e.g., robot internal states). Meanwhile, we anchor language semantics to visuomotor representations by a bidirectional cross-attention to learn contextual information for action generation, successfully overcoming the problem of semantic-physical misalignment. Extensive experiments show that CCoL achieves an average 8.0% relative improvement across three simulation suites, with up to 19.2% relative gain in human-demonstrated bimanual insertion tasks. Real-world tests on a 7-DoF robot further confirm CCoL's generalization under unseen and noisy object states.
Authors: Erwan Dereure, Robin Louiset, Laura Parkkinen, David A Menassa, David Holcman
Abstract: Parkinson's disease (PD) is a neurodegenerative disorder associated with the accumulation of misfolded alpha-synuclein aggregates, forming Lewy bodies and neuritic shape used for pathology diagnostics. Automatic analysis of immunohistochemistry histopathological images with Deep Learning provides a promising tool for better understanding the spatial organization of these aggregates. In this study, we develop an automated image processing pipeline to segment and classify these aggregates in whole-slide images (WSIs) of midbrain tissue from PD and incidental Lewy Body Disease (iLBD) cases based on weakly supervised segmentation, robust to immunohistochemical labelling variability, with a ResNet50 classifier. Our approach allows to differentiate between major aggregate morphologies, including Lewy bodies and neurites with a balanced accuracy of $80\%$. This framework paves the way for large-scale characterization of the spatial distribution and heterogeneity of alpha-synuclein aggregates in brightfield immunohistochemical tissue, and for investigating their poorly understood relationships with surrounding cells such as microglia and astrocytes.
Authors: Ziyuan Gao, Philippe Morel
Abstract: Medical AI systems face catastrophic forgetting when deployed in clinical settings, where models must learn new imaging protocols while retaining prior diagnostic capabilities. This challenge is particularly acute for medical vision-language models that must preserve complex cross-modal alignments between medical images and clinical terminology across diverse imaging modalities. We introduce Prompt- Aware Adaptive Elastic Weight Consolidation (PA-EWC), a novel continual learning approach that addresses catastrophic forgetting through prompt-guided parameter specialization. Our method systematically categorizes model parameters based on their functional roles in processing visual-descriptive, spatial-guided, and medical-semantic information, enabling targeted protection of critical knowledge while allowing adaptation to new clinical requirements. PA-EWC incorporates adaptive Fisher Information computation with gradient stability analysis and develops weighted complexity metrics based on medical terminology density. We evaluate our approach across five medical imaging datasets (Kvasir-SEG, ISIC 2018, CheXlocalize, BUSI, CAMUS) representing diverse modalities including endoscopy, dermoscopy, radiography, and ultrasound. Experimental results demonstrate that PA-EWC reduces catastrophic forgetting by up to 17.58% compared to baseline methods, with performance improvements of 4.30% on chest X-ray pathology localization and 6.06% on polyp segmentation.
Authors: Nikhil Verma, Joonas Linnosmaa, Leonardo Espinosa-Leal, Napat Vajragupta
Abstract: The paper presents the formulation, implementation, and evaluation of the ArcGD optimiser. The evaluation is conducted initially on a non-convex benchmark function and subsequently on a real-world ML dataset. The initial comparative study using the Adam optimiser is conducted on a stochastic variant of the highly non-convex and notoriously challenging Rosenbrock function, renowned for its narrow, curved valley, across dimensions ranging from 2D to 1000D and an extreme case of 50,000D. Two configurations were evaluated to eliminate learning-rate bias: (i) both using ArcGD's effective learning rate and (ii) both using Adam's default learning rate. ArcGD consistently outperformed Adam under the first setting and, although slower under the second, achieved superior final solutions in most cases. In the second evaluation, ArcGD is evaluated against state-of-the-art optimizers (Adam, AdamW, Lion, SGD) on the CIFAR-10 image classification dataset across 8 diverse MLP architectures ranging from 1 to 5 hidden layers. ArcGD achieved the highest average test accuracy (50.7%) at 20,000 iterations, outperforming AdamW (46.6%), Adam (46.8%), SGD (49.6%), and Lion (43.4%), winning or tying on 6 of 8 architectures. Notably, while Adam and AdamW showed strong early convergence at 5,000 iterations, but regressed with extended training, whereas ArcGD continued improving, demonstrating generalization and resistance to overfitting without requiring early stopping tuning. Strong performance on geometric stress tests and standard deep-learning benchmarks indicates broad applicability, highlighting the need for further exploration. Moreover, it is also shown that a limiting variant of ArcGD can be interpreted as a sign-based momentum-like update, highlighting conceptual connections between the inherent mechanisms of ArcGD and the Lion optimiser.
Authors: Ilia Larchenko, Gleb Zarin, Akash Karnatak
Abstract: We present a vision-action policy that won 1st place in the 2025 BEHAVIOR Challenge - a large-scale benchmark featuring 50 diverse long-horizon household tasks in photo-realistic simulation, requiring bimanual manipulation, navigation, and context-aware decision making. Building on the Pi0.5 architecture, we introduce several innovations. Our primary contribution is correlated noise for flow matching, which improves training efficiency and enables correlation-aware inpainting for smooth action sequences. We also apply learnable mixed-layer attention and System 2 stage tracking for ambiguity resolution. Training employs multi-sample flow matching to reduce variance, while inference uses action compression and challenge-specific correction rules. Our approach achieves 26% q-score across all 50 tasks on both public and private leaderboards.