Authors: Jiani Huang, Amish Sethi, Matthew Kuo, Mayank Keoliya, Neelay Velingker, JungHo Jung, Ser-Nam Lim, Ziyang Li, Mayur Naik
Abstract: Multi-modal large language models (MLLMs) are making rapid progress toward general-purpose embodied agents. However, current training pipelines primarily rely on high-level vision-sound-text pairs and lack fine-grained, structured alignment between pixel-level visual content and textual semantics. To overcome this challenge, we propose ESCA, a new framework for contextualizing embodied agents through structured spatial-temporal understanding. At its core is SGClip, a novel CLIP-based, open-domain, and promptable model for generating scene graphs. SGClip is trained on 87K+ open-domain videos via a neurosymbolic learning pipeline, which harnesses model-driven self-supervision from video-caption pairs and structured reasoning, thereby eliminating the need for human-labeled scene graph annotations. We demonstrate that SGClip supports both prompt-based inference and task-specific fine-tuning, excelling in scene graph generation and action localization benchmarks. ESCA with SGClip consistently improves both open-source and commercial MLLMs, achieving state-of-the-art performance across two embodied environments. Notably, it significantly reduces agent perception errors and enables open-source models to surpass proprietary baselines.
Authors: Huiming Yang
Abstract: The sparse cross-modality detector offers more advantages than its counterpart, the Bird's-Eye-View (BEV) detector, particularly in terms of adaptability for downstream tasks and computational cost savings. However, existing sparse detectors overlook the quality of token representation, leaving it with a sub-optimal foreground quality and limited performance. In this paper, we identify that the geometric structure preserved and the class distribution are the key to improving the performance of the sparse detector, and propose a Sparse Selector (SS). The core module of SS is Ray-Aware Supervision (RAS), which preserves rich geometric information during the training stage, and Class-Balanced Supervision, which adaptively reweights the salience of class semantics, ensuring that tokens associated with small objects are retained during token sampling. Thereby, outperforming other sparse multi-modal detectors in the representation of tokens. Additionally, we design Ray Positional Encoding (Ray PE) to address the distribution differences between the LiDAR modality and the image. Finally, we integrate the aforementioned module into an end-to-end sparse multi-modality detector, dubbed CrossRay3D. Experiments show that, on the challenging nuScenes benchmark, CrossRay3D achieves state-of-the-art performance with 72.4 mAP and 74.7 NDS, while running 1.84 faster than other leading methods. Moreover, CrossRay3D demonstrates strong robustness even in scenarios where LiDAR or camera data are partially or entirely missing.
Authors: Ibrahim Sheikh Mohamed, Abdullah Yahya Abdullah Omaisan
Abstract: Infrastructure in smart cities is increasingly monitored by networks of closed circuit television (CCTV) cameras. Roads, bridges and tunnels develop cracks, potholes, and fluid leaks that threaten public safety and require timely repair. Manual inspection is costly and hazardous, and existing automatic systems typically address individual defect types or provide unstructured outputs that cannot directly guide maintenance crews. This paper proposes a comprehensive pipeline that leverages street CCTV streams for multi defect detection and segmentation using the YOLO family of object detectors and passes the detections to a vision language model (VLM) for scene aware summarization. The VLM generates a structured action plan in JSON format that includes incident descriptions, recommended tools, dimensions, repair plans, and urgent alerts. We review literature on pothole, crack and leak detection, highlight recent advances in large vision language models such as QwenVL and LLaVA, and describe the design of our early prototype. Experimental evaluation on public datasets and captured CCTV clips demonstrates that the system accurately identifies diverse defects and produces coherent summaries. We conclude by discussing challenges and directions for scaling the system to city wide deployments.
Authors: Zewen Li, Zitong Yu, Qilang Ye, Weicheng Xie, Wei Zhuo, Linlin Shen
Abstract: The robust causal capability of Multimodal Large Language Models (MLLMs) hold the potential of detecting defective objects in Industrial Anomaly Detection (IAD). However, most traditional IAD methods lack the ability to provide multi-turn human-machine dialogues and detailed descriptions, such as the color of objects, the shape of an anomaly, or specific types of anomalies. At the same time, methods based on large pre-trained models have not fully stimulated the ability of large models in anomaly detection tasks. In this paper, we explore the combination of rich text semantics with both image-level and pixel-level information from images and propose IAD-GPT, a novel paradigm based on MLLMs for IAD. We employ Abnormal Prompt Generator (APG) to generate detailed anomaly prompts for specific objects. These specific prompts from the large language model (LLM) are used to activate the detection and segmentation functions of the pre-trained visual-language model (i.e., CLIP). To enhance the visual grounding ability of MLLMs, we propose Text-Guided Enhancer, wherein image features interact with normal and abnormal text prompts to dynamically select enhancement pathways, which enables language models to focus on specific aspects of visual data, enhancing their ability to accurately interpret and respond to anomalies within images. Moreover, we design a Multi-Mask Fusion module to incorporate mask as expert knowledge, which enhances the LLM's perception of pixel-level anomalies. Extensive experiments on MVTec-AD and VisA datasets demonstrate our state-of-the-art performance on self-supervised and few-shot anomaly detection and segmentation tasks, such as MVTec-AD and VisA datasets. The codes are available at \href{https://github.com/LiZeWen1225/IAD-GPT}{https://github.com/LiZeWen1225/IAD-GPT}.
URLs: https://github.com/LiZeWen1225/IAD-GPT, https://github.com/LiZeWen1225/IAD-GPT
Authors: Mahta Khoobi, Marc Sebastian von der Stueck, Felix Barajas Ordonez, Anca-Maria Iancu, Eric Corban, Julia Nowak, Aleksandar Kargaliev, Valeria Perelygina, Anna-Sophie Schott, Daniel Pinto dos Santos, Christiane Kuhl, Daniel Truhn, Sven Nebelung, Robert Siepmann
Abstract: Structured reporting (SR) and artificial intelligence (AI) may transform how radiologists interact with imaging studies. This prospective study (July to December 2024) evaluated the impact of three reporting modes: free-text (FT), structured reporting (SR), and AI-assisted structured reporting (AI-SR), on image analysis behavior, diagnostic accuracy, efficiency, and user experience. Four novice and four non-novice readers (radiologists and medical students) each analyzed 35 bedside chest radiographs per session using a customized viewer and an eye-tracking system. Outcomes included diagnostic accuracy (compared with expert consensus using Cohen's $\kappa$), reporting time per radiograph, eye-tracking metrics, and questionnaire-based user experience. Statistical analysis used generalized linear mixed models with Bonferroni post-hoc tests with a significance level of ($P \le .01$). Diagnostic accuracy was similar in FT ($\kappa = 0.58$) and SR ($\kappa = 0.60$) but higher in AI-SR ($\kappa = 0.71$, $P < .001$). Reporting times decreased from $88 \pm 38$ s (FT) to $37 \pm 18$ s (SR) and $25 \pm 9$ s (AI-SR) ($P < .001$). Saccade counts for the radiograph field ($205 \pm 135$ (FT), $123 \pm 88$ (SR), $97 \pm 58$ (AI-SR)) and total fixation duration for the report field ($11 \pm 5$ s (FT), $5 \pm 3$ s (SR), $4 \pm 1$ s (AI-SR)) were lower with SR and AI-SR ($P < .001$ each). Novice readers shifted gaze towards the radiograph in SR, while non-novice readers maintained their focus on the radiograph. AI-SR was the preferred mode. In conclusion, SR improves efficiency by guiding visual attention toward the image, and AI-prefilled SR further enhances diagnostic accuracy and user satisfaction.
Authors: Farjana Yesmin
Abstract: Machine learning models trained on imbalanced datasets often exhibit intersectional biases-systematic errors arising from the interaction of multiple attributes such as object class and environmental conditions. This paper presents a data-driven framework for analyzing and mitigating such biases in image classification. We introduce the Intersectional Fairness Evaluation Framework (IFEF), which combines quantitative fairness metrics with interpretability tools to systematically identify bias patterns in model predictions. Building on this analysis, we propose Bias-Weighted Augmentation (BWA), a novel data augmentation strategy that adapts transformation intensities based on subgroup distribution statistics. Experiments on the Open Images V7 dataset with five object classes demonstrate that BWA improves accuracy for underrepresented class-environment intersections by up to 24 percentage points while reducing fairness metric disparities by 35%. Statistical analysis across multiple independent runs confirms the significance of improvements (p < 0.05). Our methodology provides a replicable approach for analyzing and addressing intersectional biases in image classification systems.
Authors: Zia Badar
Abstract: Quantization of neural networks provides benefits of inference in less compute and memory requirements. Previous work in quantization lack two important aspects which this work provides. First almost all previous work in quantization used a non-differentiable approach and for learning; the derivative is usually set manually in backpropogation which make the learning ability of algorithm questionable, our approach is not just differentiable, we also provide proof of convergence of our approach to the optimal neural network. Second previous work in shift/logrithmic quantization either have avoided activation quantization along with weight quantization or achieved less accuracy. Learning logrithmic quantize values of form $2^n$ requires the quantization function can scale to more than 1 bit quantization which is another benifit of our quantization that it provides $n$ bits quantization as well. Our approach when tested with image classification task using imagenet dataset, resnet18 and weight quantization only achieves less than 1 percent accuracy compared to full precision accuracy while taking only 15 epochs to train using shift bit quantization and achieves comparable to SOTA approaches accuracy in both weight and activation quantization using shift bit quantization in 15 training epochs with slightly higher(only higher cpu instructions) inference cost compared to 1 bit quantization(without logrithmic quantization) and not requiring any higher precision multiplication.
Authors: Jianhan Lin, Yuchu Qin, Shuai Gao, Yikang Rui, Jie Liu, Yanjie Lv
Abstract: Well-maintained road networks are crucial for achieving Sustainable Development Goal (SDG) 11. Road surface damage not only threatens traffic safety but also hinders sustainable urban development. Accurate detection, however, remains challenging due to the diverse shapes of damages, the difficulty of capturing slender cracks with high aspect ratios, and the high error rates in small-scale damage recognition. To address these issues, we propose StripRFNet, a novel deep neural network comprising three modules: (1) a Shape Perception Module (SPM) that enhances shape discrimination via large separable kernel attention (LSKA) in multi-scale feature aggregation; (2) a Strip Receptive Field Module (SRFM) that employs large strip convolutions and pooling to capture features of slender cracks; and (3) a Small-Scale Enhancement Module (SSEM) that leverages a high-resolution P2 feature map, a dedicated detection head, and dynamic upsampling to improve small-object detection. Experiments on the RDD2022 benchmark show that StripRFNet surpasses existing methods. On the Chinese subset, it improves F1-score, mAP50, and mAP50:95 by 4.4, 2.9, and 3.4 percentage points over the baseline, respectively. On the full dataset, it achieves the highest F1-score of 80.33% compared with CRDDC'2022 participants and ORDDC'2024 Phase 2 results, while maintaining competitive inference speed. These results demonstrate that StripRFNet achieves state-of-the-art accuracy and real-time efficiency, offering a promising tool for intelligent road maintenance and sustainable infrastructure management.
Authors: Nishad Sahu (Raj), Shounak Sural (Raj), Aditya Satish Patil (Raj), Ragunathan (Raj), Rajkumar
Abstract: Reliable perception is fundamental for safety critical decision making in autonomous driving. Yet, vision based object detector neural networks remain vulnerable to uncertainty arising from issues such as data bias and distributional shifts. In this paper, we introduce ObjectTransforms, a technique for quantifying and reducing uncertainty in vision based object detection through object specific transformations at both training and inference times. At training time, ObjectTransforms perform color space perturbations on individual objects, improving robustness to lighting and color variations. ObjectTransforms also uses diffusion models to generate realistic, diverse pedestrian instances. At inference time, object perturbations are applied to detected objects and the variance of detection scores are used to quantify predictive uncertainty in real time. This uncertainty signal is then used to filter out false positives and also recover false negatives, improving the overall precision recall curve. Experiments with YOLOv8 on the NuImages 10K dataset demonstrate that our method yields notable accuracy improvements and uncertainty reduction across all object classes during training, while predicting desirably higher uncertainty values for false positives as compared to true positives during inference. Our results highlight the potential of ObjectTransforms as a lightweight yet effective mechanism for reducing and quantifying uncertainty in vision-based perception during training and inference respectively.
Authors: Chen Kong, James Fort, Aria Kang, Jonathan Wittmer, Simon Green, Tianwei Shen, Yipu Zhao, Cheng Peng, Gustavo Solaira, Andrew Berkovich, Nikhil Raina, Vijay Baiyya, Evgeniy Oleinik, Eric Huang, Fan Zhang, Julian Straub, Mark Schwesinger, Luis Pesqueira, Xiaqing Pan, Jakob Julian Engel, Carl Ren, Mingfei Yan, Richard Newcombe
Abstract: The Aria Gen 2 Pilot Dataset (A2PD) is an egocentric multimodal open dataset captured using the state-of-the-art Aria Gen 2 glasses. To facilitate timely access, A2PD is released incrementally with ongoing dataset enhancements. The initial release features Dia'ane, our primary subject, who records her daily activities alongside friends, each equipped with Aria Gen 2 glasses. It encompasses five primary scenarios: cleaning, cooking, eating, playing, and outdoor walking. In each of the scenarios, we provide comprehensive raw sensor data and output data from various machine perception algorithms. These data illustrate the device's ability to perceive the wearer, the surrounding environment, and interactions between the wearer and the environment, while maintaining robust performance across diverse users and conditions. The A2PD is publicly available at projectaria.com, with open-source tools and usage examples provided in Project Aria Tools.
Authors: Sayan Deb Sarkar, Sinisa Stekovic, Vincent Lepetit, Iro Armeni
Abstract: Transferring appearance to 3D assets using different representations of the appearance object - such as images or text - has garnered interest due to its wide range of applications in industries like gaming, augmented reality, and digital content creation. However, state-of-the-art methods still fail when the geometry between the input and appearance objects is significantly different. A straightforward approach is to directly apply a 3D generative model, but we show that this ultimately fails to produce appealing results. Instead, we propose a principled approach inspired by universal guidance. Given a pretrained rectified flow model conditioned on image or text, our training-free method interacts with the sampling process by periodically adding guidance. This guidance can be modeled as a differentiable loss function, and we experiment with two different types of guidance including part-aware losses for appearance and self-similarity. Our experiments show that our approach successfully transfers texture and geometric details to the input 3D asset, outperforming baselines both qualitatively and quantitatively. We also show that traditional metrics are not suitable for evaluating the task due to their inability of focusing on local details and comparing dissimilar inputs, in absence of ground truth data. We thus evaluate appearance transfer quality with a GPT-based system objectively ranking outputs, ensuring robust and human-like assessment, as further confirmed by our user study. Beyond showcased scenarios, our method is general and could be extended to different types of diffusion models and guidance functions.
Authors: Ahmad Arrabi, Jay hwasung Jung, J Le, A Nguyen, J Reed, E Stahl, Nathan Franssen, Scott Raymond, Safwan Wshah
Abstract: Thrombectomy is one of the most effective treatments for ischemic stroke, but it is resource and personnel-intensive. We propose employing deep learning to automate critical aspects of thrombectomy, thereby enhancing efficiency and safety. In this work, we introduce a self-supervised framework that classifies various skeletal landmarks using a regression-based pretext task. Our experiments demonstrate that our model outperforms existing methods in both regression and classification tasks. Notably, our results indicate that the positional pretext task significantly enhances downstream classification performance. Future work will focus on extending this framework toward fully autonomous C-arm control, aiming to optimize trajectories from the pelvis to the head during stroke thrombectomy procedures. All code used is available at https://github.com/AhmadArrabi/C_arm_guidance
Authors: Thanh-Huy Nguyen, Hoang-Thien Nguyen, Vi Vu, Ba-Thinh Lam, Phat Huynh, Tianyang Wang, Xingjian Li, Ulas Bagci, Min Xu
Abstract: The limited availability of annotated data in medical imaging makes semi-supervised learning increasingly appealing for its ability to learn from imperfect supervision. Recently, teacher-student frameworks have gained popularity for their training benefits and robust performance. However, jointly optimizing the entire network can hinder convergence and stability, especially in challenging scenarios. To address this for medical image segmentation, we propose DuetMatch, a novel dual-branch semi-supervised framework with asynchronous optimization, where each branch optimizes either the encoder or decoder while keeping the other frozen. To improve consistency under noisy conditions, we introduce Decoupled Dropout Perturbation, enforcing regularization across branches. We also design Pair-wise CutMix Cross-Guidance to enhance model diversity by exchanging pseudo-labels through augmented input pairs. To mitigate confirmation bias from noisy pseudo-labels, we propose Consistency Matching, refining labels using stable predictions from frozen teacher models. Extensive experiments on benchmark brain MRI segmentation datasets, including ISLES2022 and BraTS, show that DuetMatch consistently outperforms state-of-the-art methods, demonstrating its effectiveness and robustness across diverse semi-supervised segmentation scenarios.
Authors: Ahmad Arrabi, Jay Hwasung Jung, Jax Luo, Nathan Franssen, Scott Raymond, Safwan Wshah
Abstract: Accurate and reliable C-arm positioning is essential for fluoroscopy-guided interventions. However, clinical workflows rely on manual alignment that increases radiation exposure and procedural delays. In this work, we present a pipeline that autonomously navigates the C-arm to predefined anatomical landmarks utilizing X-ray images. Given an input X-ray image from an arbitrary starting location on the operating table, the model predicts a 3D displacement vector toward each target landmark along the body. To ensure reliable deployment, we capture both aleatoric and epistemic uncertainties in the model's predictions and further calibrate them using conformal prediction. The derived prediction regions are interpreted as 3D confidence regions around the predicted landmark locations. The training framework combines a probabilistic loss with skeletal pose regularization to encourage anatomically plausible outputs. We validate our approach on a synthetic X-ray dataset generated from DeepDRR. Results show not only strong localization accuracy across multiple architectures but also well-calibrated prediction bounds. These findings highlight the pipeline's potential as a component in safe and reliable autonomous C-arm systems. Code is available at https://github.com/AhmadArrabi/C_arm_guidance_APAH
Authors: Xavier Giro-i-Nieto, Nefeli Andreou, Anqi Liang, Manel Baradad, Francesc Moreno-Noguer, Aleix Martinez
Abstract: Deep generative models have shown impressive progress in recent years, making it possible to produce high quality images with a simple text prompt or a reference image. However, state of the art technology does not yet meet the quality standards offered by traditional photographic methods. For this reason, production pipelines that use generated images often include a manual stage of image quality assessment (IQA). This process is slow and expensive, especially because of the low yield of automatically generated images that pass the quality bar. The IQA workload can be reduced by introducing an automatic pre-filtering stage, that will increase the overall quality of the images sent to review and, therefore, reduce the average cost required to obtain a high quality image. We present a formula that estimates the cost savings depending on the precision and pass yield of a generic IQA engine. This formula is applied in a use case of background inpainting, showcasing a significant cost saving of 51.61% obtained with a simple AutoML solution.
Authors: Zheng Huang, Enpei Zhang, Yinghao Cai, Weikang Qiu, Carl Yang, Elynn Chen, Xiang Zhang, Rex Ying, Dawei Zhou, Yujun Yan
Abstract: Understanding how the brain encodes visual information is a central challenge in neuroscience and machine learning. A promising approach is to reconstruct visual stimuli, essentially images, from functional Magnetic Resonance Imaging (fMRI) signals. This involves two stages: transforming fMRI signals into a latent space and then using a pretrained generative model to reconstruct images. The reconstruction quality depends on how similar the latent space is to the structure of neural activity and how well the generative model produces images from that space. Yet, it remains unclear which type of latent space best supports this transformation and how it should be organized to represent visual stimuli effectively. We present two key findings. First, fMRI signals are more similar to the text space of a language model than to either a vision based space or a joint text image space. Second, text representations and the generative model should be adapted to capture the compositional nature of visual stimuli, including objects, their detailed attributes, and relationships. Building on these insights, we propose PRISM, a model that Projects fMRI sIgnals into a Structured text space as an interMediate representation for visual stimuli reconstruction. It includes an object centric diffusion module that generates images by composing individual objects to reduce object detection errors, and an attribute relationship search module that automatically identifies key attributes and relationships that best align with the neural activity. Extensive experiments on real world datasets demonstrate that our framework outperforms existing methods, achieving up to an 8% reduction in perceptual loss. These results highlight the importance of using structured text as the intermediate space to bridge fMRI signals and image reconstruction.
Authors: Mateus Pinto da Silva, Sabrina P. L. P. Correa, Hugo N. Oliveira, Ian M. Nunes, Jefersson A. dos Santos
Abstract: Mapping agriculture in tropical areas through remote sensing presents unique challenges, including the lack of high-quality annotated data, the elevated costs of labeling, data variability, and regional generalisation. This paper advocates a Data-Centric Artificial Intelligence (DCAI) perspective and pipeline, emphasizing data quality and curation as key drivers for model robustness and scalability. It reviews and prioritizes techniques such as confident learning, core-set selection, data augmentation, and active learning. The paper highlights the readiness and suitability of 25 distinct strategies in large-scale agricultural mapping pipelines. The tropical context is of high interest, since high cloudiness, diverse crop calendars, and limited datasets limit traditional model-centric approaches. This tutorial outlines practical solutions as a data-centric approach for curating and training AI models better suited to the dynamic realities of tropical agriculture. Finally, we propose a practical pipeline using the 9 most mature and straightforward methods that can be applied to a large-scale tropical agricultural mapping project.
Authors: Nyle Siddiqui, Rohit Gupta, Sirnam Swetha, Mubarak Shah
Abstract: State space models (SSMs) have emerged as a competitive alternative to transformers in various tasks. Their linear complexity and hidden-state recurrence make them particularly attractive for modeling long sequences, whereas attention becomes quadratically expensive. However, current training methods for video understanding are tailored towards transformers and fail to fully leverage the unique attributes of SSMs. For example, video models are often trained at a fixed resolution and video length to balance the quadratic scaling of attention cost against performance. Consequently, these models suffer from degraded performance when evaluated on videos with spatial and temporal resolutions unseen during training; a property we call spatio-temporal inflexibility. In the context of action recognition, this severely limits a model's ability to retain performance across both short- and long-form videos. Therefore, we propose a flexible training method that leverages and improves the inherent adaptability of SSMs. Our method samples videos at varying temporal and spatial resolutions during training and dynamically interpolates model weights to accommodate any spatio-temporal scale. This instills our SSM, which we call StretchySnake, with spatio-temporal flexibility and enables it to seamlessly handle videos ranging from short, fine-grained clips to long, complex activities. We introduce and compare five different variants of flexible training, and identify the most effective strategy for video SSMs. On short-action (UCF-101, HMDB-51) and long-action (COIN, Breakfast) benchmarks, StretchySnake outperforms transformer and SSM baselines alike by up to 28%, with strong adaptability to fine-grained actions (SSV2, Diving-48). Therefore, our method provides a simple drop-in training recipe that makes video SSMs more robust, resolution-agnostic, and efficient across diverse action recognition scenarios.
Authors: Djamel Eddine Boukhari
Abstract: Facial Beauty Prediction (FBP) is a complex and challenging computer vision task, aiming to model the subjective and intricate nature of human aesthetic perception. While deep learning models, particularly Convolutional Neural Networks (CNNs), have made significant strides, they often struggle to capture the global, holistic facial features that are critical to human judgment. Vision Transformers (ViT) address this by effectively modeling long-range spatial relationships, but their quadratic complexity can be a bottleneck. This paper introduces a novel, heterogeneous ensemble architecture, \textbf{VM-BeautyNet}, that synergistically fuses the complementary strengths of a Vision Transformer and a Mamba-based Vision model, a recent advancement in State-Space Models (SSMs). The ViT backbone excels at capturing global facial structure and symmetry, while the Mamba backbone efficiently models long-range dependencies with linear complexity, focusing on sequential features and textures. We evaluate our approach on the benchmark SCUT-FBP5500 dataset. Our proposed VM-BeautyNet achieves state-of-the-art performance, with a \textbf{Pearson Correlation (PC) of 0.9212}, a \textbf{Mean Absolute Error (MAE) of 0.2085}, and a \textbf{Root Mean Square Error (RMSE) of 0.2698}. Furthermore, through Grad-CAM visualizations, we provide interpretability analysis that confirms the complementary feature extraction of the two backbones, offering new insights into the model's decision-making process and presenting a powerful new architectural paradigm for computational aesthetics.
Authors: Vishal Manikanden, Aniketh Bandlamudi, Daniel Haehn
Abstract: Oral Cavity Squamous Cell Carcinoma (OCSCC) is the most common type of head and neck cancer. Due to the subtle nature of its early stages, deep and hidden areas of development, and slow growth, OCSCC often goes undetected, leading to preventable deaths. However, properly trained Convolutional Neural Networks (CNNs), with their precise image segmentation techniques and ability to apply kernel matrices to modify the RGB values of images for accurate image pattern recognition, would be an effective means for early detection of OCSCC. Pairing this neural network with image capturing and processing hardware would allow increased efficacy in OCSCC detection. The aim of our project is to develop a Convolutional Neural Network trained to recognize OCSCC, as well as to design a physical hardware system to capture and process detailed images, in order to determine the image quality required for accurate predictions. A CNN was trained on 4293 training images consisting of benign and malignant tumors, as well as negative samples, and was evaluated for its precision, recall, and Mean Average Precision (mAP) in its predictions of OCSCC. A testing dataset of randomly assorted images of cancerous, non-cancerous, and negative images was chosen, and each image was altered to represent 5 common resolutions. This test data set was thoroughly analyzed by the CNN and predictions were scored on the basis of accuracy. The designed enhancement hardware was used to capture detailed images, and its impact was scored. An application was developed to facilitate the testing process and bring open access to the CNN. Images of increasing resolution resulted in higher-accuracy predictions on a logarithmic scale, demonstrating the diminishing returns of higher pixel counts.
Authors: Claire McLean, Makenzie Meendering, Tristan Swartz, Orri Gabbay, Alexandra Olsen, Rachel Jacobs, Nicholas Rosen, Philippe de Bree, Tony Garcia, Gadsden Merrill, Jake Sandakly, Julia Buffalini, Neham Jain, Steven Krenn, Moneish Kumar, Dejan Markovic, Evonne Ng, Fabian Prada, Andrew Saba, Siwei Zhang, Vasu Agrawal, Tim Godisart, Alexander Richard, Michael Zollhoefer
Abstract: The Codec Avatars Lab at Meta introduces Embody 3D, a multimodal dataset of 500 individual hours of 3D motion data from 439 participants collected in a multi-camera collection stage, amounting to over 54 million frames of tracked 3D motion. The dataset features a wide range of single-person motion data, including prompted motions, hand gestures, and locomotion; as well as multi-person behavioral and conversational data like discussions, conversations in different emotional states, collaborative activities, and co-living scenarios in an apartment-like space. We provide tracked human motion including hand tracking and body shape, text annotations, and a separate audio track for each participant.
Authors: Baicheng Li, Zike Yan, Dong Wu, Hongbin Zha
Abstract: Human behaviors are the major causes of scene dynamics and inherently contain rich cues regarding the dynamics. This paper formalizes a new task of proactive scene decomposition and reconstruction, an online approach that leverages human-object interactions to iteratively disassemble and reconstruct the environment. By observing these intentional interactions, we can dynamically refine the decomposition and reconstruction process, addressing inherent ambiguities in static object-level reconstruction. The proposed system effectively integrates multiple tasks in dynamic environments such as accurate camera and object pose estimation, instance decomposition, and online map updating, capitalizing on cues from human-object interactions in egocentric live streams for a flexible, progressive alternative to conventional object-level reconstruction methods. Aided by the Gaussian splatting technique, accurate and consistent dynamic scene modeling is achieved with photorealistic and efficient rendering. The efficacy is validated in multiple real-world scenarios with promising advantages.
Authors: Yue Zheng, Xiufang Shi, Jiming Chen, Yuanchao Shu
Abstract: Video anomaly detection (VAD) has rapidly advanced by recent development of Vision-Language Models (VLMs). While these models offer superior zero-shot detection capabilities, their immense computational cost and unstable visual grounding performance hinder real-time deployment. To overcome these challenges, we introduce Cerberus, a two-stage cascaded system designed for efficient yet accurate real-time VAD. Cerberus learns normal behavioral rules offline, and combines lightweight filtering with fine-grained VLM reasoning during online inference. The performance gains of Cerberus come from two key innovations: motion mask prompting and rule-based deviation detection. The former directs the VLM's attention to regions relevant to motion, while the latter identifies anomalies as deviations from learned norms rather than enumerating possible anomalies. Extensive evaluations on four datasets show that Cerberus on average achieves 57.68 fps on an NVIDIA L40S GPU, a 151.79$\times$ speedup, and 97.2\% accuracy comparable to the state-of-the-art VLM-based VAD methods, establishing it as a practical solution for real-time video analytics.
Authors: Ryoto Miyamoto, Xin Fan, Fuyuko Kido, Tsuneo Matsumoto, Hayato Yamana
Abstract: OpenLVLM-MIA is a new benchmark that highlights fundamental challenges in evaluating membership inference attacks (MIA) against large vision-language models (LVLMs). While prior work has reported high attack success rates, our analysis suggests that these results often arise from detecting distributional bias introduced during dataset construction rather than from identifying true membership status. To address this issue, we introduce a controlled benchmark of 6{,}000 images where the distributions of member and non-member samples are carefully balanced, and ground-truth membership labels are provided across three distinct training stages. Experiments using OpenLVLM-MIA demonstrated that the performance of state-of-the-art MIA methods converged to random chance under unbiased conditions. By offering a transparent and unbiased benchmark, OpenLVLM-MIA clarifies the current limitations of MIA research on LVLMs and provides a solid foundation for developing stronger privacy-preserving techniques.
Authors: Rui Yang, Huining Li, Yiyi Long, Xiaojun Wu, Shengfeng He
Abstract: Generating sketches guided by reference styles requires precise transfer of stroke attributes, such as line thickness, deformation, and texture sparsity, while preserving semantic structure and content fidelity. To this end, we propose Stroke2Sketch, a novel training-free framework that introduces cross-image stroke attention, a mechanism embedded within self-attention layers to establish fine-grained semantic correspondences and enable accurate stroke attribute transfer. This allows our method to adaptively integrate reference stroke characteristics into content images while maintaining structural integrity. Additionally, we develop adaptive contrast enhancement and semantic-focused attention to reinforce content preservation and foreground emphasis. Stroke2Sketch effectively synthesizes stylistically faithful sketches that closely resemble handcrafted results, outperforming existing methods in expressive stroke control and semantic coherence. Codes are available at https://github.com/rane7/Stroke2Sketch.
Authors: Wenhao Wang, Longqi Cai, Taihong Xiao, Yuxiao Wang, Ming-Hsuan Yang
Abstract: This paper presents a systematic study of scaling laws for the deepfake detection task. Specifically, we analyze the model performance against the number of real image domains, deepfake generation methods, and training images. Since no existing dataset meets the scale requirements for this research, we construct ScaleDF, the largest dataset to date in this field, which contains over 5.8 million real images from 51 different datasets (domains) and more than 8.8 million fake images generated by 102 deepfake methods. Using ScaleDF, we observe power-law scaling similar to that shown in large language models (LLMs). Specifically, the average detection error follows a predictable power-law decay as either the number of real domains or the number of deepfake methods increases. This key observation not only allows us to forecast the number of additional real domains or deepfake methods required to reach a target performance, but also inspires us to counter the evolving deepfake technology in a data-centric manner. Beyond this, we examine the role of pre-training and data augmentations in deepfake detection under scaling, as well as the limitations of scaling itself.
Authors: Yuyao Zhang, Yu-Wing Tai
Abstract: Ultra-high-resolution text-to-image generation demands both fine-grained texture synthesis and globally coherent structure, yet current diffusion models remain constrained to sub-$1K \times 1K$ resolutions due to the prohibitive quadratic complexity of attention and the scarcity of native $4K$ training data. We present \textbf{Scale-DiT}, a new diffusion framework that introduces hierarchical local attention with low-resolution global guidance, enabling efficient, scalable, and semantically coherent image synthesis at ultra-high resolutions. Specifically, high-resolution latents are divided into fixed-size local windows to reduce attention complexity from quadratic to near-linear, while a low-resolution latent equipped with scaled positional anchors injects global semantics. A lightweight LoRA adaptation bridges global and local pathways during denoising, ensuring consistency across structure and detail. To maximize inference efficiency, we repermute token sequence in Hilbert curve order and implement a fused-kernel for skipping masked operations, resulting in a GPU-friendly design. Extensive experiments demonstrate that Scale-DiT achieves more than $2\times$ faster inference and lower memory usage compared to dense attention baselines, while reliably scaling to $4K \times 4K$ resolution without requiring additional high-resolution training data. On both quantitative benchmarks (FID, IS, CLIP Score) and qualitative comparisons, Scale-DiT delivers superior global coherence and sharper local detail, matching or outperforming state-of-the-art methods that rely on native 4K training. Taken together, these results highlight hierarchical local attention with guided low-resolution anchors as a promising and effective approach for advancing ultra-high-resolution image generation.
Authors: Yi Wei (College of Information Science,Electronic Engineering, Zhejiang University, Hangzhou, China), Shunpu Tang (College of Information Science,Electronic Engineering, Zhejiang University, Hangzhou, China), Liang Zhao (College of Information Science,Electronic Engineering, Zhejiang University, Hangzhou, China), Qiangian Yang (College of Information Science,Electronic Engineering, Zhejiang University, Hangzhou, China)
Abstract: Recent advances in diffusion models have driven remarkable progress in image generation. However, the generation process remains computationally intensive, and users often need to iteratively refine prompts to achieve the desired results, further increasing latency and placing a heavy burden on cloud resources. To address this challenge, we propose DiffusionX, a cloud-edge collaborative framework for efficient multi-round, prompt-based generation. In this system, a lightweight on-device diffusion model interacts with users by rapidly producing preview images, while a high-capacity cloud model performs final refinements after the prompt is finalized. We further introduce a noise level predictor that dynamically balances the computation load, optimizing the trade-off between latency and cloud workload. Experiments show that DiffusionX reduces average generation time by 15.8% compared with Stable Diffusion v1.5, while maintaining comparable image quality. Moreover, it is only 0.9% slower than Tiny-SD with significantly improved image quality, thereby demonstrating efficiency and scalability with minimal overhead.
Authors: Haiyue Sun, Qingdong He, Jinlong Peng, Peng Tang, Jiangning Zhang, Junwei Zhu, Xiaobin Hu, Shuicheng Yan
Abstract: Autoregressive Model (AR) has shown remarkable success in conditional image generation. However, these approaches for multiple reference generation struggle with decoupling different reference identities. In this work, we propose the TokenAR framework, specifically focused on a simple but effective token-level enhancement mechanism to address reference identity confusion problem. Such token-level enhancement consists of three parts, 1). Token Index Embedding clusters the tokens index for better representing the same reference images; 2). Instruct Token Injection plays as a role of extra visual feature container to inject detailed and complementary priors for reference tokens; 3). The identity-token disentanglement strategy (ITD) explicitly guides the token representations toward independently representing the features of each identity.This token-enhancement framework significantly augments the capabilities of existing AR based methods in conditional image generation, enabling good identity consistency while preserving high quality background reconstruction. Driven by the goal of high-quality and high-diversity in multi-subject generation, we introduce the InstructAR Dataset, the first open-source, large-scale, multi-reference input, open domain image generation dataset that includes 28K training pairs, each example has two reference subjects, a relative prompt and a background with mask annotation, curated for multiple reference image generation training and evaluating. Comprehensive experiments validate that our approach surpasses current state-of-the-art models in multiple reference image generation task. The implementation code and datasets will be made publicly. Codes are available, see https://github.com/lyrig/TokenAR
Authors: Junha Song, Sangdoo Yun, Dongyoon Han, Jaegul Choo, Byeongho Heo
Abstract: A dominant assumption in Multimodal Language Model (MLLM) research is that its performance is largely inherited from the LLM backbone, given its immense parameter scale and remarkable capabilities. This has created a void in the understanding of the vision encoder, which determines how MLLMs perceive images. The recent shift in MLLM training paradigms, from Supervised Finetuning (SFT) to Reinforcement Learning (RL), magnifies this oversight-namely, the significant lack of analysis on how such training reshapes the vision encoder as well as the MLLM. To address this, we first investigate the impact of training strategies on MLLMs, where RL shows a clear advantage over SFT in strongly vision-related VQA benchmarks. Motivated by this, we conduct a critical yet under-explored analysis of the vision encoder of MLLMs through diverse and in-depth experiments, ranging from ImageNet classification and segmentation to gradient visualization. Our results demonstrate that MLLM's post-training strategy (i.e., SFT or RL) not only leads to distinct outcomes on MLLM downstream tasks, but also fundamentally reshapes MLLM's underlying visual representations. Specifically, the key finding of our study is that RL produces stronger and precisely localized visual representations compared to SFT, boosting the ability of the vision encoder for MLLM. We then reframe our findings into a simple recipe for building strong vision encoders for MLLMs, Preference-Instructed Vision OpTimization (PIVOT). When integrated into MLLMs, a PIVOT-trained vision encoder outperforms even larger and more heavily-trained counterparts, despite requiring less than 1% of the computational cost of standard vision pretraining. This result opens an effective and efficient path for advancing the vision backbones of MLLMs. Project page available at https://june-page.github.io/pivot/
Authors: Bo Peng, Jie Lu, Guangquan Zhang, Zhen Fang
Abstract: This paper investigates the recently emerged problem of Language-assisted Image Clustering (LaIC), where textual semantics are leveraged to improve the discriminability of visual representations to facilitate image clustering. Due to the unavailability of true class names, one of core challenges of LaIC lies in how to filter positive nouns, i.e., those semantically close to the images of interest, from unlabeled wild corpus data. Existing filtering strategies are predominantly based on the off-the-shelf feature space learned by CLIP; however, despite being intuitive, these strategies lack a rigorous theoretical foundation. To fill this gap, we propose a novel gradient-based framework, termed as GradNorm, which is theoretically guaranteed and shows strong empirical performance. In particular, we measure the positiveness of each noun based on the magnitude of gradients back-propagated from the cross-entropy between the predicted target distribution and the softmax output. Theoretically, we provide a rigorous error bound to quantify the separability of positive nouns by GradNorm and prove that GradNorm naturally subsumes existing filtering strategies as extremely special cases of itself. Empirically, extensive experiments show that GradNorm achieves the state-of-the-art clustering performance on various benchmarks.
Authors: Pulin Li, Guocheng Wu, Li Yin, Yuxin Zheng, Wei Zhang, Yanjie Zhou
Abstract: Social manufacturing leverages community collaboration and scattered resources to realize mass individualization in modern industry. However, this paradigm shift also introduces substantial challenges in quality control, particularly in defect detection. The main difficulties stem from three aspects. First, products often have highly customized configurations. Second, production typically involves fragmented, small-batch orders. Third, imaging environments vary considerably across distributed sites. To overcome the scarcity of real-world datasets and tailored algorithms, we introduce the Mass Individualization Robust Anomaly Detection (MIRAD) dataset. As the first benchmark explicitly designed for anomaly detection in social manufacturing, MIRAD captures three critical dimensions of this domain: (1) diverse individualized products with large intra-class variation, (2) data collected from six geographically dispersed manufacturing nodes, and (3) substantial imaging heterogeneity, including variations in lighting, background, and motion conditions. We then conduct extensive evaluations of state-of-the-art (SOTA) anomaly detection methods on MIRAD, covering one-class, multi-class, and zero-shot approaches. Results show a significant performance drop across all models compared with conventional benchmarks, highlighting the unresolved complexities of defect detection in real-world individualized production. By bridging industrial requirements and academic research, MIRAD provides a realistic foundation for developing robust quality control solutions essential for Industry 5.0. The dataset is publicly available at https://github.com/wu33learn/MIRAD.
Authors: Mohammad Javad Ahmadi, Iman Gandomi, Parisa Abdi, Seyed-Farzad Mohammadi, Amirhossein Taslimi, Mehdi Khodaparast, Hassan Hashemi, Mahdi Tavakoli, Hamid D. Taghirad
Abstract: The development of computer-assisted surgery systems depends on large-scale, annotated datasets. Current resources for cataract surgery often lack the diversity and annotation depth needed to train generalizable deep-learning models. To address this gap, we present a dataset of 3,000 phacoemulsification cataract surgery videos from two surgical centers, performed by surgeons with a range of experience levels. This resource is enriched with four annotation layers: temporal surgical phases, instance segmentation of instruments and anatomical structures, instrument-tissue interaction tracking, and quantitative skill scores based on the established competency rubrics like the ICO-OSCAR. The technical quality of the dataset is supported by a series of benchmarking experiments for key surgical AI tasks, including workflow recognition, scene segmentation, and automated skill assessment. Furthermore, we establish a domain adaptation baseline for the phase recognition task by training a model on a subset of surgical centers and evaluating its performance on a held-out center. The dataset and annotations are available in Google Form (https://docs.google.com/forms/d/e/1FAIpQLSfmyMAPSTGrIy2sTnz0-TMw08ZagTimRulbAQcWdaPwDy187A/viewform?usp=dialog).
Authors: Rishi Raj Sahoo, Surbhi Saswati Mohanty, Subhankar Mishra
Abstract: Road potholes pose significant safety hazards and maintenance challenges, particularly on India's diverse and under-maintained road networks. This paper presents iWatchRoadv2, a fully automated end-to-end platform for real-time pothole detection, GPS-based geotagging, and dynamic road health visualization using OpenStreetMap (OSM). We curated a self-annotated dataset of over 7,000 dashcam frames capturing diverse Indian road conditions, weather patterns, and lighting scenarios, which we used to fine-tune the Ultralytics YOLO model for accurate pothole detection. The system synchronizes OCR-extracted video timestamps with external GPS logs to precisely geolocate each detected pothole, enriching detections with comprehensive metadata, including road segment attribution and contractor information managed through an optimized backend database. iWatchRoadv2 introduces intelligent governance features that enable authorities to link road segments with contract metadata through a secure login interface. The system automatically sends alerts to contractors and officials when road health deteriorates, supporting automated accountability and warranty enforcement. The intuitive web interface delivers actionable analytics to stakeholders and the public, facilitating evidence-driven repair planning, budget allocation, and quality assessment. Our cost-effective and scalable solution streamlines frame processing and storage while supporting seamless public engagement for urban and rural deployments. By automating the complete pothole monitoring lifecycle, from detection to repair verification, iWatchRoadv2 enables data-driven smart city management, transparent governance, and sustainable improvements in road infrastructure maintenance. The platform and live demonstration are accessible at https://smlab.niser.ac.in/project/iwatchroad.
Authors: Tianhang Cheng, Albert J. Zhai, Evan Z. Chen, Rui Zhou, Yawen Deng, Zitong Li, Kejie Zhao, Janice Shiu, Qianyu Zhao, Yide Xu, Xinlei Wang, Yuan Shen, Sheng Wang, Lisa Ainsworth, Kaiyu Guan, Shenlong Wang
Abstract: Learning 3D parametric shape models of objects has gained popularity in vision and graphics and has showed broad utility in 3D reconstruction, generation, understanding, and simulation. While powerful models exist for humans and animals, equally expressive approaches for modeling plants are lacking. In this work, we present Demeter, a data-driven parametric model that encodes key factors of a plant morphology, including topology, shape, articulation, and deformation into a compact learned representation. Unlike previous parametric models, Demeter handles varying shape topology across various species and models three sources of shape variation: articulation, subcomponent shape variation, and non-rigid deformation. To advance crop plant modeling, we collected a large-scale, ground-truthed dataset from a soybean farm as a testbed. Experiments show that Demeter effectively synthesizes shapes, reconstructs structures, and simulates biophysical processes. Code and data is available at https://tianhang-cheng.github.io/Demeter/.
Authors: Yeh Keng Hao, Hsu Tzu Wei, Sun Min
Abstract: With the increasing ubiquity of AR/VR devices, the deployment of deep learning models on edge devices has become a critical challenge. These devices require real-time inference, low power consumption, and minimal latency. Many framework designers face the conundrum of balancing efficiency and performance. We design a light framework that adopts an encoder-decoder architecture and introduces several key contributions aimed at improving both efficiency and accuracy. We apply sparse convolution on a ResNet-18 backbone to exploit the inherent sparsity in hand pose images, achieving a 42% end-to-end efficiency improvement. Moreover, we propose our SPLite decoder. This new architecture significantly boosts the decoding process's frame rate by 3.1x on the Raspberry Pi 5, while maintaining accuracy on par. To further optimize performance, we apply quantization-aware training, reducing memory usage while preserving accuracy (PA-MPJPE increases only marginally from 9.0 mm to 9.1 mm on FreiHAND). Overall, our system achieves a 2.98x speed-up on a Raspberry Pi 5 CPU (BCM2712 quad-core Arm A76 processor). Our method is also evaluated on compound benchmark datasets, demonstrating comparable accuracy to state-of-the-art approaches while significantly enhancing computational efficiency.
Authors: Changyue Shi, Minghao Chen, Yiping Mao, Chuxiao Yang, Xinyuan Hu, Jiajun Ding, Zhou Yu
Abstract: Bridging the gap between complex human instructions and precise 3D object grounding remains a significant challenge in vision and robotics. Existing 3D segmentation methods often struggle to interpret ambiguous, reasoning-based instructions, while 2D vision-language models that excel at such reasoning lack intrinsic 3D spatial understanding. In this paper, we introduce REALM, an innovative MLLM-agent framework that enables open-world reasoning-based segmentation without requiring extensive 3D-specific post-training. We perform segmentation directly on 3D Gaussian Splatting representations, capitalizing on their ability to render photorealistic novel views that are highly suitable for MLLM comprehension. As directly feeding one or more rendered views to the MLLM can lead to high sensitivity to viewpoint selection, we propose a novel Global-to-Local Spatial Grounding strategy. Specifically, multiple global views are first fed into the MLLM agent in parallel for coarse-level localization, aggregating responses to robustly identify the target object. Then, several close-up novel views of the object are synthesized to perform fine-grained local segmentation, yielding accurate and consistent 3D masks. Extensive experiments show that REALM achieves remarkable performance in interpreting both explicit and implicit instructions across LERF, 3D-OVS, and our newly introduced REALM3D benchmarks. Furthermore, our agent framework seamlessly supports a range of 3D interaction tasks, including object removal, replacement, and style transfer, demonstrating its practical utility and versatility. Project page: https://ChangyueShi.github.io/REALM.
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: Aidyn Ubingazhibov, R\'emi Pautrat, Iago Su\'arez, Shaohui Liu, Marc Pollefeys, Viktor Larsson
Abstract: Lines and points are complementary local features, whose combination has proven effective for applications such as SLAM and Structure-from-Motion. The backbone of these pipelines are the local feature matchers, establishing correspondences across images. Traditionally, point and line matching have been treated as independent tasks. Recently, GlueStick proposed a GNN-based network that simultaneously operates on points and lines to establish matches. While running a single joint matching reduced the overall computational complexity, the heavy architecture prevented real-time applications or deployment to edge devices. Inspired by recent progress in point matching, we propose LightGlueStick, a lightweight matcher for points and line segments. The key novel component in our architecture is the Attentional Line Message Passing (ALMP), which explicitly exposes the connectivity of the lines to the network, allowing for efficient communication between nodes. In thorough experiments we show that LightGlueStick establishes a new state-of-the-art across different benchmarks. The code is available at https://github.com/aubingazhib/LightGlueStick.
Authors: Haoran Sun, Chen Cai, Huiping Zhuang, Kong Aik Lee, Lap-Pui Chau, Yi Wang
Abstract: The rapid development of deepfake video technology has not only facilitated artistic creation but also made it easier to spread misinformation. Traditional deepfake video detection (DVD) methods face issues such as a lack of transparency in their principles and insufficient generalization capabilities to cope with evolving forgery techniques. This highlights an urgent need for detectors that can identify forged content and provide verifiable reasoning explanations. This paper proposes the explainable deepfake video detection (EDVD) task and designs the EDVD-LLaMA multimodal, a large language model (MLLM) reasoning framework, which provides traceable reasoning processes alongside accurate detection results and trustworthy explanations. Our approach first incorporates a Spatio-Temporal Subtle Information Tokenization (ST-SIT) to extract and fuse global and local cross-frame deepfake features, providing rich spatio-temporal semantic information input for MLLM reasoning. Second, we construct a Fine-grained Multimodal Chain-of-Thought (Fg-MCoT) mechanism, which introduces facial feature data as hard constraints during the reasoning process to achieve pixel-level spatio-temporal video localization, suppress hallucinated outputs, and enhance the reliability of the chain of thought. In addition, we build an Explainable Reasoning FF++ benchmark dataset (ER-FF++set), leveraging structured data to annotate videos and ensure quality control, thereby supporting dual supervision for reasoning and detection. Extensive experiments demonstrate that EDVD-LLaMA achieves outstanding performance and robustness in terms of detection accuracy, explainability, and its ability to handle cross-forgery methods and cross-dataset scenarios. Compared to previous DVD methods, it provides a more explainable and superior solution. The source code and dataset will be publicly available.
Authors: Kunyu Peng, Di Wen, Jia Fu, Jiamin Wu, Kailun Yang, Junwei Zheng, Ruiping Liu, Yufan Chen, Yuqian Fu, Danda Pani Paudel, Luc Van Gool, Rainer Stiefelhagen
Abstract: Referring Atomic Video Action Recognition (RAVAR) aims to recognize fine-grained, atomic-level actions of a specific person of interest conditioned on natural language descriptions. Distinct from conventional action recognition and detection tasks, RAVAR emphasizes precise language-guided action understanding, which is particularly critical for interactive human action analysis in complex multi-person scenarios. In this work, we extend our previously introduced RefAVA dataset to RefAVA++, which comprises >2.9 million frames and >75.1k annotated persons in total. We benchmark this dataset using baselines from multiple related domains, including atomic action localization, video question answering, and text-video retrieval, as well as our earlier model, RefAtomNet. Although RefAtomNet surpasses other baselines by incorporating agent attention to highlight salient features, its ability to align and retrieve cross-modal information remains limited, leading to suboptimal performance in localizing the target person and predicting fine-grained actions. To overcome the aforementioned limitations, we introduce RefAtomNet++, a novel framework that advances cross-modal token aggregation through a multi-hierarchical semantic-aligned cross-attention mechanism combined with multi-trajectory Mamba modeling at the partial-keyword, scene-attribute, and holistic-sentence levels. In particular, scanning trajectories are constructed by dynamically selecting the nearest visual spatial tokens at each timestep for both partial-keyword and scene-attribute levels. Moreover, we design a multi-hierarchical semantic-aligned cross-attention strategy, enabling more effective aggregation of spatial and temporal tokens across different semantic hierarchies. Experiments show that RefAtomNet++ establishes new state-of-the-art results. The dataset and code are released at https://github.com/KPeng9510/refAVA2.
Authors: Chien Thai, Mai Xuan Trang, Huong Ninh, Hoang Hiep Ly, Anh Son Le
Abstract: Detecting rotated objects accurately and efficiently is a significant challenge in computer vision, particularly in applications such as aerial imagery, remote sensing, and autonomous driving. Although traditional object detection frameworks are effective for axis-aligned objects, they often underperform in scenarios involving rotated objects due to their limitations in capturing orientation variations. This paper introduces an improved loss function aimed at enhancing detection accuracy and robustness by leveraging the Gaussian bounding box representation and Bhattacharyya distance. In addition, we advocate for the use of an anisotropic Gaussian representation to address the issues associated with isotropic variance in square-like objects. Our proposed method addresses these challenges by incorporating a rotation-invariant loss function that effectively captures the geometric properties of rotated objects. We integrate this proposed loss function into state-of-the-art deep learning-based rotated object detection detectors, and extensive experiments demonstrated significant improvements in mean Average Precision metrics compared to existing methods. The results highlight the potential of our approach to establish new benchmark in rotated object detection, with implications for a wide range of applications requiring precise and reliable object localization irrespective of orientation.
Authors: Jaekyun Park, Hye Won Chung
Abstract: In the era of large-scale foundation models, fully fine-tuning pretrained networks for each downstream task is often prohibitively resource-intensive. Prompt tuning offers a lightweight alternative by introducing tunable prompts while keeping the backbone frozen. However, existing visual prompt tuning methods often fail to specialize the prompts or enrich the representation space--especially when applied to self-supervised backbones. We show that these limitations become especially pronounced in challenging tasks and data-scarce settings, where effective adaptation is most critical. In this work, we introduce VIPAMIN, a visual prompt initialization strategy that enhances adaptation of self-supervised models by (1) aligning prompts with semantically informative regions in the embedding space, and (2) injecting novel representational directions beyond the pretrained subspace. Despite its simplicity--requiring only a single forward pass and lightweight operations--VIPAMIN consistently improves performance across diverse tasks and dataset sizes, setting a new state of the art in visual prompt tuning. Our code is available at https://github.com/iamjaekyun/vipamin.
Authors: Shan Xiong, Jiabao Chen, Ye Wang, Jialin Peng
Abstract: Annotation-efficient segmentation of the numerous mitochondria instances from various electron microscopy (EM) images is highly valuable for biological and neuroscience research. Although unsupervised domain adaptation (UDA) methods can help mitigate domain shifts and reduce the high costs of annotating each domain, they typically have relatively low performance in practical applications. Thus, we investigate weakly supervised domain adaptation (WDA) that utilizes additional sparse point labels on the target domain, which require minimal annotation effort and minimal expert knowledge. To take full use of the incomplete and imprecise point annotations, we introduce a multitask learning framework that jointly conducts segmentation and center detection with a novel cross-teaching mechanism and class-focused cross-domain contrastive learning. While leveraging unlabeled image regions is essential, we introduce segmentation self-training with a novel instance-aware pseudo-label (IPL) selection strategy. Unlike existing methods that typically rely on pixel-wise pseudo-label filtering, the IPL semantically selects reliable and diverse pseudo-labels with the help of the detection task. Comprehensive validations and comparisons on challenging datasets demonstrate that our method outperforms existing UDA and WDA methods, significantly narrowing the performance gap with the supervised upper bound. Furthermore, under the UDA setting, our method also achieves substantial improvements over other UDA techniques.
Authors: Peiran Xu, Xicheng Gong, Yadong MU
Abstract: In this work we concentrate on the task of goal-oriented Vision-and-Language Navigation (VLN). Existing methods often make decisions based on historical information, overlooking the future implications and long-term outcomes of the actions. In contrast, we aim to develop a foresighted agent. Specifically, we draw upon Q-learning to train a Q-model using large-scale unlabeled trajectory data, in order to learn the general knowledge regarding the layout and object relations within indoor scenes. This model can generate a Q-feature, analogous to the Q-value in traditional Q-network, for each candidate action, which describes the potential future information that may be observed after taking the specific action. Subsequently, a cross-modal future encoder integrates the task-agnostic Q-feature with navigation instructions to produce a set of action scores reflecting future prospects. These scores, when combined with the original scores based on history, facilitate an A*-style searching strategy to effectively explore the regions that are more likely to lead to the destination. Extensive experiments conducted on widely used goal-oriented VLN datasets validate the effectiveness of the proposed method.
Authors: Haocheng Tang, Ruoke Yan, Xinhui Yin, Qi Zhang, Xinfeng Zhang, Siwei Ma, Wen Gao, Chuanmin Jia
Abstract: Recent advances in 3D Gaussian Splatting (3DGS) have enabled fast, photorealistic rendering of dynamic 3D scenes, showing strong potential in immersive communication. However, in digital human encoding and transmission, the compression methods based on general 3DGS representations are limited by the lack of human priors, resulting in suboptimal bitrate efficiency and reconstruction quality at the decoder side, which hinders their application in streamable 3D avatar systems. We propose HGC-Avatar, a novel Hierarchical Gaussian Compression framework designed for efficient transmission and high-quality rendering of dynamic avatars. Our method disentangles the Gaussian representation into a structural layer, which maps poses to Gaussians via a StyleUNet-based generator, and a motion layer, which leverages the SMPL-X model to represent temporal pose variations compactly and semantically. This hierarchical design supports layer-wise compression, progressive decoding, and controllable rendering from diverse pose inputs such as video sequences or text. Since people are most concerned with facial realism, we incorporate a facial attention mechanism during StyleUNet training to preserve identity and expression details under low-bitrate constraints. Experimental results demonstrate that HGC-Avatar provides a streamable solution for rapid 3D avatar rendering, while significantly outperforming prior methods in both visual quality and compression efficiency.
Authors: Lukas Selch, Yufang Hou, M. Jehanzeb Mirza, Sivan Doveh, James Glass, Rogerio Feris, Wei Lin
Abstract: Large Multimodal Models (LMMs) are increasingly applied to scientific research, yet it remains unclear whether they can reliably understand and reason over the multimodal complexity of papers. A central challenge lies in detecting and resolving inconsistencies across text, figures, tables, and equations, issues that are often subtle, domain-specific, and ultimately undermine clarity, reproducibility, and trust. Existing benchmarks overlook this issue, either isolating single modalities or relying on synthetic errors that fail to capture real-world complexity. We introduce PRISMM-Bench (Peer-Review-sourced Inconsistency Set for Multimodal Models), the first benchmark grounded in real reviewer-flagged inconsistencies in scientific papers. Through a multi-stage pipeline of review mining, LLM-assisted filtering and human verification, we curate 262 inconsistencies from 242 papers. Based on this set, we design three tasks, namely inconsistency identification, remedy and pair matching, which assess a model's capacity to detect, correct, and reason over inconsistencies across different modalities. Furthermore, to address the notorious problem of choice-only shortcuts in multiple-choice evaluation, where models exploit answer patterns without truly understanding the question, we further introduce structured JSON-based answer representations that minimize linguistic biases by reducing reliance on superficial stylistic cues. We benchmark 21 leading LMMs, including large open-weight models (GLM-4.5V 106B, InternVL3 78B) and proprietary models (Gemini 2.5 Pro, GPT-5 with high reasoning). Results reveal strikingly low performance (26.1-54.2%), underscoring the challenge of multimodal scientific reasoning and motivating progress towards trustworthy scientific assistants.
Authors: Franko \v{S}iki\'c, Sven Lon\v{c}ari\'c
Abstract: Out-of-stock (OOS) detection is a very important retail verification process that aims to infer the unavailability of products in their designated areas on the shelf. In this paper, we introduce OOS-DSD, a novel deep learning-based method that advances OOS detection through auxiliary learning. In particular, we extend a well-established YOLOv8 object detection architecture with additional convolutional branches to simultaneously detect OOS, segment products, and estimate scene depth. While OOS detection and product segmentation branches are trained using ground truth data, the depth estimation branch is trained using pseudo-labeled annotations produced by the state-of-the-art (SOTA) depth estimation model Depth Anything V2. Furthermore, since the aforementioned pseudo-labeled depth estimates display relative depth, we propose an appropriate depth normalization procedure that stabilizes the training process. The experimental results show that the proposed method surpassed the performance of the SOTA OOS detection methods by 1.8% of the mean average precision (mAP). In addition, ablation studies confirm the effectiveness of auxiliary learning and the proposed depth normalization procedure, with the former increasing mAP by 3.7% and the latter by 4.2%.
Authors: Duygu Sap, Martin Lotz, Connor Mattinson
Abstract: We propose a method for image categorization and retrieval that leverages graphs and a graph attention network (GAT)-based autoencoder. Our approach is representative-centric, that is, we execute the categorization and retrieval process via the representative models we construct for the images and image categories. We utilize a graph where nodes represent images (or their representatives) and edges capture similarity relationships. GAT highlights important features and relationships between images, enabling the autoencoder to construct context-aware latent representations that capture the key features of each image relative to its neighbors. We obtain category representatives from these embeddings and categorize a query image by comparing its representative to the category representatives. We then retrieve the most similar image to the query image within its identified category. We demonstrate the effectiveness of our representative-centric approach through experiments with both the GAT autoencoders and standard feature-based techniques.
Authors: Jihoon Kwon, Kyle Min, Jy-yong Sohn
Abstract: Despite recent advances, vision-language models trained with standard contrastive objectives still struggle with compositional reasoning -- the ability to understand structured relationships between visual and linguistic elements. This shortcoming is largely due to the tendency of the text encoder to focus on individual words rather than their relations, a limitation reinforced by contrastive training that primarily aligns words with visual objects. In this paper, we introduce REconstruction and Alignment of text Descriptions (READ), a fine-tuning method designed to enhance compositional reasoning by adding two auxiliary objectives to the contrastive learning: (1) a token-level reconstruction objective, where a frozen pre-trained decoder reconstructs alternative captions based on the embedding of the original caption; and (2) a sentence-level alignment objective, which explicitly aligns paraphrased sentences in the embedding space. We show that READ-CLIP, a model derived by applying the READ method to the pre-trained CLIP model, achieves the state-of-the-art performance across five major compositional reasoning benchmarks, outperforming the strongest conventional fine-tuning baseline by up to 4.1%. Furthermore, applying the READ to existing CLIP variants (including NegCLIP and FSC-CLIP) also improves performance on these benchmarks. Quantitative and qualitative analyses reveal that our proposed objectives -- reconstruction and alignment -- offer complementary benefits: the former encourages the encoder to capture relationships between words within a caption, while the latter ensures consistent representations for paraphrases expressed with different wording.
Authors: Binyuan Huang, Yongdong Luo, Xianda Guo, Xiawu Zheng, Zheng Zhu, Jiahui Pan, Chengju Zhou
Abstract: Deep learning-based gait recognition has achieved great success in various applications. The key to accurate gait recognition lies in considering the unique and diverse behavior patterns in different motion regions, especially when covariates affect visual appearance. However, existing methods typically use predefined regions for temporal modeling, with fixed or equivalent temporal scales assigned to different types of regions, which makes it difficult to model motion regions that change dynamically over time and adapt to their specific patterns. To tackle this problem, we introduce a Region-aware Dynamic Aggregation and Excitation framework (GaitRDAE) that automatically searches for motion regions, assigns adaptive temporal scales and applies corresponding attention. Specifically, the framework includes two core modules: the Region-aware Dynamic Aggregation (RDA) module, which dynamically searches the optimal temporal receptive field for each region, and the Region-aware Dynamic Excitation (RDE) module, which emphasizes the learning of motion regions containing more stable behavior patterns while suppressing attention to static regions that are more susceptible to covariates. Experimental results show that GaitRDAE achieves state-of-the-art performance on several benchmark datasets.
Authors: Guangyu Lin, Li Lin, Christina P. Walker, Daniel S. Schiff, Shu Hu
Abstract: The rapid proliferation of AI-generated content, driven by advances in generative adversarial networks, diffusion models, and multimodal large language models, has made the creation and dissemination of synthetic media effortless, heightening the risks of misinformation, particularly political deepfakes that distort truth and undermine trust in political institutions. In turn, governments, research institutions, and industry have strongly promoted deepfake detection initiatives as solutions. Yet, most existing models are trained and validated on synthetic, laboratory-controlled datasets, limiting their generalizability to the kinds of real-world political deepfakes circulating on social platforms that affect the public. In this work, we introduce the first systematic benchmark based on the Political Deepfakes Incident Database, a curated collection of real-world political deepfakes shared on social media since 2018. Our study includes a systematic evaluation of state-of-the-art deepfake detectors across academia, government, and industry. We find that the detectors from academia and government perform relatively poorly. While paid detection tools achieve relatively higher performance than free-access models, all evaluated detectors struggle to generalize effectively to authentic political deepfakes, and are vulnerable to simple manipulations, especially in the video domain. Results urge the need for politically contextualized deepfake detection frameworks to better safeguard the public in real-world settings.
Authors: Yiyang Huang, Liang Shi, Yitian Zhang, Yi Xu, Yun Fu
Abstract: Large Vision-Language Models (LVLMs) excel in diverse cross-modal tasks. However, object hallucination, where models produce plausible but inaccurate object descriptions, remains a significant challenge. In contrast to previous work focusing on LLM components, this paper is the first to trace LVLM hallucinations to visual encoders and identifies three key issues: statistical bias, inherent bias, and vulnerability. To address these challenges, we propose SHIELD, a training-free framework that mitigates hallucinations through three strategies: re-weighting visual tokens to reduce statistical bias, introducing noise-derived tokens to counter inherent bias, and applying adversarial attacks with contrastive decoding to address vulnerability. Experiments demonstrate that SHIELD effectively mitigates object hallucinations across diverse benchmarks and LVLM families. Moreover, SHIELD achieves strong performance on the general LVLM benchmark, highlighting its broad applicability. Code will be released.
Authors: Jiaying Zhu, Yurui Zhu, Xin Lu, Wenrui Yan, Dong Li, Kunlin Liu, Xueyang Fu, Zheng-Jun Zha
Abstract: Multimodal Large Language Models (MLLMs) encounter significant computational and memory bottlenecks from the massive number of visual tokens generated by high-resolution images or multi-image inputs. Previous token compression techniques are often constrained by heuristic rules that risk discarding critical information. They may suffer from biases, such as attention sinks, that lead to sharp performance drops under aggressive compression ratios. To address these limitations, we reformulate token compression as a lightweight plug-and-play framework that reformulates token compression into an end-to-end learnable decision process. To be specific, we propose VisionSelector, a scorer module decoupled from the MLLM backbone that incorporates a differentiable Top-K mechanism and a curriculum annealing strategy to bridge the training-inference gap, enabling efficient and adaptive token selection various arbitrary compression rates. Remarkably lightweight with only 12.85M trainable parameters, VisionSelector demonstrates generalization across various compression rates and adaptively identifying critical tokens. This leads to superior performance across all compression budgets, evidenced by preserving 100% accuracy on MME with 30% retention budget, outperforming prior methods by 12.14% at 10% retention budget, and doubling prefill speed. Our code is available at https://github.com/JulietChoo/VisionSelector .
Authors: Melika Filvantorkaman, Maral Filvan Torkaman
Abstract: Medical imaging plays a vital role in modern diagnostics; however, interpreting high-resolution radiological data remains time-consuming and susceptible to variability among clinicians. Traditional image processing techniques often lack the precision, robustness, and speed required for real-time clinical use. To overcome these limitations, this paper introduces a deep learning framework for real-time medical image analysis designed to enhance diagnostic accuracy and computational efficiency across multiple imaging modalities, including X-ray, CT, and MRI. The proposed system integrates advanced neural network architectures such as U-Net, EfficientNet, and Transformer-based models with real-time optimization strategies including model pruning, quantization, and GPU acceleration. The framework enables flexible deployment on edge devices, local servers, and cloud infrastructures, ensuring seamless interoperability with clinical systems such as PACS and EHR. Experimental evaluations on public benchmark datasets demonstrate state-of-the-art performance, achieving classification accuracies above 92%, segmentation Dice scores exceeding 91%, and inference times below 80 milliseconds. Furthermore, visual explanation tools such as Grad-CAM and segmentation overlays enhance transparency and clinical interpretability. These results indicate that the proposed framework can substantially accelerate diagnostic workflows, reduce clinician workload, and support trustworthy AI integration in time-critical healthcare environments.
Authors: Sebastian Mocanu, Emil Slusanschi, Marius Leordeanu
Abstract: This paper presents a vision-only autonomous flight system for small UAVs operating in controlled indoor environments. The system combines semantic segmentation with monocular depth estimation to enable obstacle avoidance, scene exploration, and autonomous safe landing operations without requiring GPS or expensive sensors such as LiDAR. A key innovation is an adaptive scale factor algorithm that converts non-metric monocular depth predictions into accurate metric distance measurements by leveraging semantic ground plane detection and camera intrinsic parameters, achieving a mean distance error of 14.4 cm. The approach uses a knowledge distillation framework where a color-based Support Vector Machine (SVM) teacher generates training data for a lightweight U-Net student network (1.6M parameters) capable of real-time semantic segmentation. For more complex environments, the SVM teacher can be replaced with a state-of-the-art segmentation model. Testing was conducted in a controlled 5x4 meter laboratory environment with eight cardboard obstacles simulating urban structures. Extensive validation across 30 flight tests in a real-world environment and 100 flight tests in a digital-twin environment demonstrates that the combined segmentation and depth approach increases the distance traveled during surveillance and reduces mission time while maintaining 100% success rates. The system is further optimized through end-to-end learning, where a compact student neural network learns complete flight policies from demonstration data generated by our best-performing method, achieving an 87.5% autonomous mission success rate. This work advances practical vision-based drone navigation in structured environments, demonstrating solutions for metric depth estimation and computational efficiency challenges that enable deployment on resource-constrained platforms.
Authors: Young-Jun Lee, Byung-Kwan Lee, Jianshu Zhang, Yechan Hwang, Byungsoo Ko, Han-Gyu Kim, Dongyu Yao, Xuankun Rong, Eojin Joo, Seung-Ho Han, Bowon Ko, Ho-Jin Choi
Abstract: Vision-and-Language Models (VLMs) have shown impressive capabilities on single-turn benchmarks, yet real-world applications often demand more intricate multi-turn dialogues. Existing multi-turn datasets (e.g, MMDU, ConvBench) only partially capture the breadth and depth of conversational scenarios encountered by users. In this work, we introduce MultiVerse, a novel multi-turn conversation benchmark featuring 647 dialogues - each averaging four turns - derived from a diverse set of 12 popular VLM evaluation benchmarks. With 484 tasks and 484 interaction goals, MultiVerse covers a wide range of topics, from factual knowledge and perception to advanced reasoning tasks such as mathematics and coding. To facilitate robust assessment, we propose a checklist-based evaluation method that leverages GPT-4o as the automated evaluator, measuring performance across 37 key aspects, including perceptual accuracy, linguistic clarity, and factual correctness. We evaluate 18 VLMs on MultiVerse, revealing that even the strongest models (e.g., GPT-4o) achieve only a 50% success rate in complex multi-turn conversations, highlighting the dataset's challenging nature. Notably, we find that providing full dialogue context significantly enhances performance for smaller or weaker models, emphasizing the importance of in-context learning. We believe MultiVerse is a landscape of evaluating multi-turn interaction abilities for VLMs.
Authors: Aaron Ray, Jacob Arkin, Harel Biggie, Chuchu Fan, Luca Carlone, Nicholas Roy
Abstract: In order to provide a robot with the ability to understand and react to a user's natural language inputs, the natural language must be connected to the robot's underlying representations of the world. Recently, large language models (LLMs) and 3D scene graphs (3DSGs) have become a popular choice for grounding natural language and representing the world. In this work, we address the challenge of using LLMs with 3DSGs to ground natural language. Existing methods encode the scene graph as serialized text within the LLM's context window, but this encoding does not scale to large or rich 3DSGs. Instead, we propose to use a form of Retrieval Augmented Generation to select a subset of the 3DSG relevant to the task. We encode a 3DSG in a graph database and provide a query language interface (Cypher) as a tool to the LLM with which it can retrieve relevant data for language grounding. We evaluate our approach on instruction following and scene question-answering tasks and compare against baseline context window and code generation methods. Our results show that using Cypher as an interface to 3D scene graphs scales significantly better to large, rich graphs on both local and cloud-based models. This leads to large performance improvements in grounded language tasks while also substantially reducing the token count of the scene graph content. A video supplement is available at https://www.youtube.com/watch?v=zY_YI9giZSA.
Authors: Yuntian Wang, Xilin Yang, Che-Yung Shen, Nir Pillar, Aydogan Ozcan
Abstract: We introduce Universal and Transferable Adversarial Perturbations (UTAP) for pathology foundation models that reveal critical vulnerabilities in their capabilities. Optimized using deep learning, UTAP comprises a fixed and weak noise pattern that, when added to a pathology image, systematically disrupts the feature representation capabilities of multiple pathology foundation models. Therefore, UTAP induces performance drops in downstream tasks that utilize foundation models, including misclassification across a wide range of unseen data distributions. In addition to compromising the model performance, we demonstrate two key features of UTAP: (1) universality: its perturbation can be applied across diverse field-of-views independent of the dataset that UTAP was developed on, and (2) transferability: its perturbation can successfully degrade the performance of various external, black-box pathology foundation models - never seen before. These two features indicate that UTAP is not a dedicated attack associated with a specific foundation model or image dataset, but rather constitutes a broad threat to various emerging pathology foundation models and their applications. We systematically evaluated UTAP across various state-of-the-art pathology foundation models on multiple datasets, causing a significant drop in their performance with visually imperceptible modifications to the input images using a fixed noise pattern. The development of these potent attacks establishes a critical, high-standard benchmark for model robustness evaluation, highlighting a need for advancing defense mechanisms and potentially providing the necessary assets for adversarial training to ensure the safe and reliable deployment of AI in pathology.
Authors: Christopher Thirgood, Oscar Mendez, Erin Ling, Jon Storey, Simon Hadfield
Abstract: Hyperspectral images (HSI) promise to support a range of new applications in computer vision. Recent research has explored the feasibility of generalizable Spectral Reconstruction (SR), the problem of recovering a HSI from a natural three-channel color image in unseen scenarios. However, previous Multi-Scale Attention (MSA) works have only demonstrated sufficient generalizable results for very sparse spectra, while modern HSI sensors contain hundreds of channels. This paper introduces a novel approach to spectral reconstruction via our HYbrid knowledge Distillation and spectral Reconstruction Architecture (HYDRA). Using a Teacher model that encapsulates latent hyperspectral image data and a Student model that learns mappings from natural images to the Teacher's encoded domain, alongside a novel training method, we achieve high-quality spectral reconstruction. This addresses key limitations of prior SR models, providing SOTA performance across all metrics, including an 18\% boost in accuracy, and faster inference times than current SOTA models at various channel depths.
Authors: Yejie Guo, Yunzhong Hou, Wufei Ma, Meng Tang, Ming-Hsuan Yang
Abstract: Spatial reasoning, the ability to ground language in 3D understanding, remains a persistent challenge for Vision-Language Models (VLMs). We identify two fundamental bottlenecks: inadequate 3D understanding capabilities stemming from 2D-centric pre-training, and reasoning failures induced by redundant 3D information. To address these, we first construct a Minimal Sufficient Set (MSS) of information before answering a given question: a compact selection of 3D perception results from \textit{expert models}. We introduce MSSR (Minimal Sufficient Spatial Reasoner), a dual-agent framework that implements this principle. A Perception Agent programmatically queries 3D scenes using a versatile perception toolbox to extract sufficient information, including a novel SOG (Situated Orientation Grounding) module that robustly extracts language-grounded directions. A Reasoning Agent then iteratively refines this information to pursue minimality, pruning redundant details and requesting missing ones in a closed loop until the MSS is curated. Extensive experiments demonstrate that our method, by explicitly pursuing both sufficiency and minimality, significantly improves accuracy and achieves state-of-the-art performance across two challenging benchmarks. Furthermore, our framework produces interpretable reasoning paths, offering a promising source of high-quality training data for future models. Source code is available at https://github.com/gyj155/mssr.
Authors: Huy Minh Nhat Nguyen, Triet Hoang Minh Dao, Chau Vinh Hoang Truong, Cuong Tuan Nguyen
Abstract: Optical Coherence Tomography (OCT) is a widely used non-invasive imaging technique that provides detailed three-dimensional views of the retina, which are essential for the early and accurate diagnosis of ocular diseases. Consequently, OCT image analysis and processing have emerged as key research areas in biomedical imaging. However, acquiring paired datasets of clean and real-world noisy OCT images for supervised denoising models remains a formidable challenge due to intrinsic speckle noise and practical constraints in clinical imaging environments. To address these issues, we propose SDPA++: A General Framework for Self-Supervised Denoising with Patch Aggregation. Our novel approach leverages only noisy OCT images by first generating pseudo-ground-truth images through self-fusion and self-supervised denoising. These refined images then serve as targets to train an ensemble of denoising models using a patch-based strategy that effectively enhances image clarity. Performance improvements are validated via metrics such as Contrast-to-Noise Ratio (CNR), Mean Square Ratio (MSR), Texture Preservation (TP), and Edge Preservation (EP) on the real-world dataset from the IEEE SPS Video and Image Processing Cup. Notably, the VIP Cup dataset contains only real-world noisy OCT images without clean references, highlighting our method's potential for improving image quality and diagnostic outcomes in clinical practice.
Authors: Tianxin Wei, Yifan Chen, Xinrui He, Wenxuan Bao, Jingrui He
Abstract: Distribution shifts between training and testing samples frequently occur in practice and impede model generalization performance. This crucial challenge thereby motivates studies on domain generalization (DG), which aim to predict the label on unseen target domain data by solely using data from source domains. It is intuitive to conceive the class-separated representations learned in contrastive learning (CL) are able to improve DG, while the reality is quite the opposite: users observe directly applying CL deteriorates the performance. We analyze the phenomenon with the insights from CL theory and discover lack of intra-class connectivity in the DG setting causes the deficiency. We thus propose a new paradigm, domain-connecting contrastive learning (DCCL), to enhance the conceptual connectivity across domains and obtain generalizable representations for DG. On the data side, more aggressive data augmentation and cross-domain positive samples are introduced to improve intra-class connectivity. On the model side, to better embed the unseen test domains, we propose model anchoring to exploit the intra-class connectivity in pre-trained representations and complement the anchoring with generative transformation loss. Extensive experiments on five standard DG benchmarks are performed. The results verify that DCCL outperforms state-of-the-art baselines even without domain supervision. The detailed model implementation and the code are provided through https://github.com/weitianxin/DCCL
Authors: Liu Haojie, Gao Suixiang
Abstract: We present HumanCM, a one-step human motion prediction framework built upon consistency models. Instead of relying on multi-step denoising as in diffusion-based methods, HumanCM performs efficient single-step generation by learning a self-consistent mapping between noisy and clean motion states. The framework adopts a Transformer-based spatiotemporal architecture with temporal embeddings to model long-range dependencies and preserve motion coherence. Experiments on Human3.6M and HumanEva-I demonstrate that HumanCM achieves comparable or superior accuracy to state-of-the-art diffusion models while reducing inference steps by up to two orders of magnitude.
Authors: Xiongkun Linghu, Jiangyong Huang, Ziyu Zhu, Baoxiong Jia, Siyuan Huang
Abstract: Existing research on 3D Large Language Models (LLMs) still struggles to achieve grounded question-answering, primarily due to the under-exploration of the mech- anism of human-like scene-object grounded reasoning. This paper bridges the gap by presenting a novel framework. We first introduce a grounded Chain-of- Thought reasoning method in 3D scenes (SCENECOT), decoupling a complex reasoning task into simpler and manageable problems, and building corresponding visual clues based on multimodal expert modules. To enable such a method, we develop SCENECOT-185K, the first large-scale grounded CoT reasoning dataset, consisting of 185K high-quality instances. Extensive experiments across various complex 3D scene reasoning benchmarks demonstrate that our new framework achieves strong performance with high grounding-QA coherence. To the best of our knowledge, this is the first successful application of CoT reasoning to 3D scene understanding, enabling step-by-step human-like reasoning and showing potential for extension to broader 3D scene understanding scenarios.
Authors: Jianbiao Mei, Yu Yang, Xuemeng Yang, Licheng Wen, Jiajun Lv, Botian Shi, Yong Liu
Abstract: End-to-end autonomous driving systems increasingly rely on vision-centric world models to understand and predict their environment. However, a common ineffectiveness in these models is the full reconstruction of future scenes, which expends significant capacity on redundantly modeling static backgrounds. To address this, we propose IR-WM, an Implicit Residual World Model that focuses on modeling the current state and evolution of the world. IR-WM first establishes a robust bird's-eye-view representation of the current state from the visual observation. It then leverages the BEV features from the previous timestep as a strong temporal prior and predicts only the "residual", i.e., the changes conditioned on the ego-vehicle's actions and scene context. To alleviate error accumulation over time, we further apply an alignment module to calibrate semantic and dynamic misalignments. Moreover, we investigate different forecasting-planning coupling schemes and demonstrate that the implicit future state generated by world models substantially improves planning accuracy. On the nuScenes benchmark, IR-WM achieves top performance in both 4D occupancy forecasting and trajectory planning.
Authors: Tianyang Dou, Ming Li, Jiangying Qin, Xuan Liao, Jiageng Zhong, Armin Gruen, Mengyi Deng
Abstract: Coral reefs are vital yet fragile ecosystems that require accurate large-scale mapping for effective conservation. Although global products such as the Allen Coral Atlas provide unprecedented coverage of global coral reef distri-bution, their predictions are frequently limited in spatial precision and semantic consistency, especially in regions requiring fine-grained boundary delineation. To address these challenges, we propose UKANFormer, a novel se-mantic segmentation model designed to achieve high-precision mapping under noisy supervision derived from Allen Coral Atlas. Building upon the UKAN architecture, UKANFormer incorporates a Global-Local Transformer (GL-Trans) block in the decoder, enabling the extraction of both global semantic structures and local boundary details. In experiments, UKANFormer achieved a coral-class IoU of 67.00% and pixel accuracy of 83.98%, outperforming conventional baselines under the same noisy labels setting. Remarkably, the model produces predictions that are visually and structurally more accurate than the noisy labels used for training. These results challenge the notion that data quality directly limits model performance, showing that architectural design can mitigate label noise and sup-port scalable mapping under imperfect supervision. UKANFormer provides a foundation for ecological monitoring where reliable labels are scarce.
Authors: Xinqing Li, Xin He, Le Zhang, Yun Liu
Abstract: Embodied AI requires agents that perceive, act, and anticipate how actions reshape future world states. World models serve as internal simulators that capture environment dynamics, enabling forward and counterfactual rollouts to support perception, prediction, and decision making. This survey presents a unified framework for world models in embodied AI. Specifically, we formalize the problem setting and learning objectives, and propose a three-axis taxonomy encompassing: (1) Functionality, Decision-Coupled vs. General-Purpose; (2) Temporal Modeling, Sequential Simulation and Inference vs. Global Difference Prediction; (3) Spatial Representation, Global Latent Vector, Token Feature Sequence, Spatial Latent Grid, and Decomposed Rendering Representation. We systematize data resources and metrics across robotics, autonomous driving, and general video settings, covering pixel prediction quality, state-level understanding, and task performance. Furthermore, we offer a quantitative comparison of state-of-the-art models and distill key open challenges, including the scarcity of unified datasets and the need for evaluation metrics that assess physical consistency over pixel fidelity, the trade-off between model performance and the computational efficiency required for real-time control, and the core modeling difficulty of achieving long-horizon temporal consistency while mitigating error accumulation. Finally, we maintain a curated bibliography at https://github.com/Li-Zn-H/AwesomeWorldModels.
Authors: Erik Riise, Mehmet Onurcan Kaya, Dim P. Papadopoulos
Abstract: While inference-time scaling through search has revolutionized Large Language Models, translating these gains to image generation has proven difficult. Recent attempts to apply search strategies to continuous diffusion models show limited benefits, with simple random sampling often performing best. We demonstrate that the discrete, sequential nature of visual autoregressive models enables effective search for image generation. We show that beam search substantially improves text-to-image generation, enabling a 2B parameter autoregressive model to outperform a 12B parameter diffusion model across benchmarks. Systematic ablations show that this advantage comes from the discrete token space, which allows early pruning and computational reuse, and our verifier analysis highlights trade-offs between speed and reasoning capability. These findings suggest that model architecture, not just scale, is critical for inference-time optimization in visual generation.
Authors: Ivan Molodetskikh, Kirill Malyshev, Mark Mirgaleev, Nikita Zagainov, Evgeney Bogatyrev, Dmitriy Vatolin
Abstract: Generative image super-resolution (SR) is rapidly advancing in visual quality and detail restoration. As the capacity of SR models expands, however, so does their tendency to produce artifacts: incorrect, visually disturbing details that reduce perceived quality. Crucially, their perceptual impact varies: some artifacts are barely noticeable while others strongly degrade the image. We argue that artifacts should be characterized by their prominence to human observers rather than treated as uniform binary defects. Motivated by this, we present a novel dataset of 1302 artifact examples from 11 contemporary image-SR methods, where each artifact is paired with a crowdsourced prominence score. Building on this dataset, we train a lightweight regressor that produces spatial prominence heatmaps and outperforms existing methods at detecting prominent artifacts. We release the dataset and code to facilitate prominence-aware evaluation and mitigation of SR artifacts.
Authors: Shengyu Zhu, Fan, Fuxuan Zhang
Abstract: Image restoration is a fundamental and challenging task in computer vision, where CNN-based frameworks demonstrate significant computational efficiency. However, previous CNN-based methods often face challenges in adequately restoring fine texture details, which are limited by the small receptive field of CNN structures and the lack of channel feature modeling. In this paper, we propose WaMaIR, which is a novel framework with a large receptive field for image perception and improves the reconstruction of texture details in restored images. Specifically, we introduce the Global Multiscale Wavelet Transform Convolutions (GMWTConvs) for expandding the receptive field to extract image features, preserving and enriching texture features in model inputs. Meanwhile, we propose the Mamba-Based Channel-Aware Module (MCAM), explicitly designed to capture long-range dependencies within feature channels, which enhancing the model sensitivity to color, edges, and texture information. Additionally, we propose Multiscale Texture Enhancement Loss (MTELoss) for image restoration to guide the model in preserving detailed texture structures effectively. Extensive experiments confirm that WaMaIR outperforms state-of-the-art methods, achieving better image restoration and efficient computational performance of the model.
Authors: Thuy Phuong Vu, Dinh-Cuong Hoang, Minhhuy Le, Phan Xuan Tan
Abstract: Recent research has made significant progress in localizing and editing image regions based on text. However, most approaches treat these regions in isolation, relying solely on local cues without accounting for how each part contributes to the overall visual and semantic composition. This often results in inconsistent edits, unnatural transitions, or loss of coherence across the image. In this work, we propose Region in Context, a novel framework for text-conditioned image editing that performs multilevel semantic alignment between vision and language, inspired by the human ability to reason about edits in relation to the whole scene. Our method encourages each region to understand its role within the global image context, enabling precise and harmonized changes. At its core, the framework introduces a dual-level guidance mechanism: regions are represented with full-image context and aligned with detailed region-level descriptions, while the entire image is simultaneously matched to a comprehensive scene-level description generated by a large vision-language model. These descriptions serve as explicit verbal references of the intended content, guiding both local modifications and global structure. Experiments show that it produces more coherent and instruction-aligned results. Code is available at: https://github.com/thuyvuphuong/Region-in-Context.git
Authors: Mingzheng Zhang, Jinfeng Gao, Dan Xu, Jiangrui Yu, Yuhan Qiao, Lan Chen, Jin Tang, Xiao Wang
Abstract: X-ray image-based medical report generation (MRG) is a pivotal area in artificial intelligence that can significantly reduce diagnostic burdens for clinicians and patient wait times. Existing MRG models predominantly rely on Large Language Models (LLMs) to improve report generation, with limited exploration of pre-trained vision foundation models or advanced fine-tuning techniques. Mainstream frameworks either avoid fine-tuning or utilize simplistic methods like LoRA, often neglecting the potential of enhancing cross-attention mechanisms. Additionally, while Transformer-based models dominate vision-language tasks, non-Transformer architectures, such as the Mamba network, remain underexplored for medical report generation, presenting a promising avenue for future research. In this paper, we propose EMRRG, a novel X-ray report generation framework that fine-tunes pre-trained Mamba networks using parameter-efficient methods. Specifically, X-ray images are divided into patches, tokenized, and processed by an SSM-based vision backbone for feature extraction, with Partial LoRA yielding optimal performance. An LLM with a hybrid decoder generates the medical report, enabling end-to-end training and achieving strong results on benchmark datasets. Extensive experiments on three widely used benchmark datasets fully validated the effectiveness of our proposed strategies for the X-ray MRG. The source code of this paper will be released on https://github.com/Event-AHU/Medical_Image_Analysis.
Authors: Junbo Li, Weimin Yuan, Yinuo Wang, Yue Zeng, Shihao Shu, Cai Meng, Xiangzhi Bai
Abstract: Accurate 6D pose estimation of 3D objects is a fundamental task in computer vision, and current research typically predicts the 6D pose by establishing correspondences between 2D image features and 3D model features. However, these methods often face difficulties with textureless objects and varying illumination conditions. To overcome these limitations, we propose GS2POSE, a novel approach for 6D object pose estimation. GS2POSE formulates a pose regression algorithm inspired by the principles of Bundle Adjustment (BA). By leveraging Lie algebra, we extend the capabilities of 3DGS to develop a pose-differentiable rendering pipeline, which iteratively optimizes the pose by comparing the input image to the rendered image. Additionally, GS2POSE updates color parameters within the 3DGS model, enhancing its adaptability to changes in illumination. Compared to previous models, GS2POSE demonstrates accuracy improvements of 1.4\%, 2.8\% and 2.5\% on the T-LESS, LineMod-Occlusion and LineMod datasets, respectively.
Authors: Shihao Ji, Zihui Song
Abstract: The remarkable zero-shot reasoning capabilities of large-scale Visual Language Models (VLMs) on static images have yet to be fully translated to the video domain. Conventional video understanding models often rely on extensive, task-specific training on annotated datasets, a process that is both costly and limited in scalability. This paper introduces a novel, training-free framework for video understanding that circumvents end-to-end training by synergistically combining the rich semantic priors of pre-trained VLMs with classic machine learning algorithms for pattern discovery. Our core idea is to reframe video understanding as a self-supervised spatio-temporal clustering problem within a high-dimensional semantic feature space. The proposed pipeline first transforms a video stream into a semantic feature trajectory using the frozen visual encoder of a pre-trained VLM. Subsequently, we employ Kernel Temporal Segmentation (KTS), a robust machine learning technique, to partition the continuous feature stream into discrete, semantically coherent event segments. These segments are then subjected to unsupervised density-based clustering to identify recurring macroscopic scenes and themes throughout the video. By selecting representative keyframes from each discovered cluster and leveraging the VLM's generative capabilities for textual description, our framework automatically produces a structured, multi-modal summary of the video content. This approach provides an effective, interpretable, and model-agnostic pathway for zero-shot, automated structural analysis of video content.
Authors: Jiazhen Liu, Long Chen
Abstract: Integrating diverse visual capabilities into a unified model is a significant trend in Multimodal Large Language Models (MLLMs). Among these, the inclusion of segmentation poses a distinct set of challenges. To equip MLLMs with pixel-level segmentation abilities, prevailing methods require finetuning the model to produce specific outputs compatible with a mask decoder. This process typically alters the model's output space and compromises its intrinsic generalization, which undermines the goal of building a unified model. We introduce LENS (Leveraging kEypoiNts for MLLMs' Segmentation), a novel plug-and-play solution. LENS attaches a lightweight, trainable head to a completely frozen MLLM. By refining the spatial cues embedded in attention maps, LENS extracts keypoints and describes them into point-wise features directly compatible with the mask decoder. Extensive experiments validate our approach: LENS achieves segmentation performance competitive with or superior to that of retraining-based methods. Crucially, it does so while fully preserving the MLLM's generalization capabilities, which are significantly degraded by finetuning approaches. As such, the attachable design of LENS establishes an efficient and powerful paradigm for extending MLLMs, paving the way for truly multi-talented, unified models.
Authors: Sara Hatami Rostami, Behrooz Nasihatkon
Abstract: This paper presents a fully unsupervised approach for binary road segmentation (road vs. non-road), eliminating the reliance on costly manually labeled datasets. The method leverages scene geometry and temporal cues to distinguish road from non-road regions. Weak labels are first generated from geometric priors, marking pixels above the horizon as non-road and a predefined quadrilateral in front of the vehicle as road. In a refinement stage, temporal consistency is enforced by tracking local feature points across frames and penalizing inconsistent label assignments using mutual information maximization. This enhances both precision and temporal stability. On the Cityscapes dataset, the model achieves an Intersection-over-Union (IoU) of 0.82, demonstrating high accuracy with a simple design. These findings demonstrate the potential of combining geometric constraints and temporal consistency for scalable unsupervised road segmentation in autonomous driving.
Authors: Chengxuan Zhu, Shuchen Weng, Jiacong Fang, Peixuan Zhang, Si Li, Chao Xu, Boxin Shi
Abstract: Photographic style, as a composition of certain photographic concepts, is the charm behind renowned photographers. But learning and transferring photographic style need a profound understanding of how the photo is edited from the unknown original appearance. Previous works either fail to learn meaningful photographic concepts from reference images, or cannot preserve the content of the content image. To tackle these issues, we proposed a Personalized Image Filter (PIF). Based on a pretrained text-to-image diffusion model, the generative prior enables PIF to learn the average appearance of photographic concepts, as well as how to adjust them according to text prompts. PIF then learns the photographic style of reference images with the textual inversion technique, by optimizing the prompts for the photographic concepts. PIF shows outstanding performance in extracting and transferring various kinds of photographic style. Project page: https://pif.pages.dev/
URLs: https://pif.pages.dev/
Authors: Zhenpeng Zhang, Yi Wang, Shanglei Chai, Yingying Liu, Zekai Xie, Wenhao Huang, Pengyu Li, Zipei Luo, Dajiang Lu, Yibin Tian
Abstract: Lychee is a high-value subtropical fruit. The adoption of vision-based harvesting robots can significantly improve productivity while reduce reliance on labor. High-quality data are essential for developing such harvesting robots. However, there are currently no consistently and comprehensively annotated open-source lychee datasets featuring fruits in natural growing environments. To address this, we constructed a dataset to facilitate lychee detection and maturity classification. Color (RGB) images were acquired under diverse weather conditions, and at different times of the day, across multiple lychee varieties, such as Nuomici, Feizixiao, Heiye, and Huaizhi. The dataset encompasses three different ripeness stages and contains 11,414 images, consisting of 878 raw RGB images, 8,780 augmented RGB images, and 1,756 depth images. The images are annotated with 9,658 pairs of lables for lychee detection and maturity classification. To improve annotation consistency, three individuals independently labeled the data, and their results were then aggregated and verified by a fourth reviewer. Detailed statistical analyses were done to examine the dataset. Finally, we performed experiments using three representative deep learning models to evaluate the dataset. It is publicly available for academic
Authors: Yahia Battach, Abdulwahab Felemban, Faizan Farooq Khan, Yousef A. Radwan, Xiang Li, Fabio Marchese, Sara Beery, Burton H. Jones, Francesca Benzoni, Mohamed Elhoseiny
Abstract: Coral reefs are rapidly declining due to anthropogenic pressures such as climate change, underscoring the urgent need for scalable, automated monitoring. We introduce ReefNet, a large public coral reef image dataset with point-label annotations mapped to the World Register of Marine Species (WoRMS). ReefNet aggregates imagery from 76 curated CoralNet sources and an additional site from Al Wajh in the Red Sea, totaling approximately 925000 genus-level hard coral annotations with expert-verified labels. Unlike prior datasets, which are often limited by size, geography, or coarse labels and are not ML-ready, ReefNet offers fine-grained, taxonomically mapped labels at a global scale to WoRMS. We propose two evaluation settings: (i) a within-source benchmark that partitions each source's images for localized evaluation, and (ii) a cross-source benchmark that withholds entire sources to test domain generalization. We analyze both supervised and zero-shot classification performance on ReefNet and find that while supervised within-source performance is promising, supervised performance drops sharply across domains, and performance is low across the board for zero-shot models, especially for rare and visually similar genera. This provides a challenging benchmark intended to catalyze advances in domain generalization and fine-grained coral classification. We will release our dataset, benchmarking code, and pretrained models to advance robust, domain-adaptive, global coral reef monitoring and conservation.
Authors: Abdur Rahman, Mohammad Marufuzzaman, Jason Street, Haifeng Wang, Veera G. Gude, Randy Buchanan
Abstract: Accurate and quick prediction of wood chip moisture content is critical for optimizing biofuel production and ensuring energy efficiency. The current widely used direct method (oven drying) is limited by its longer processing time and sample destructiveness. On the other hand, existing indirect methods, including near-infrared spectroscopy-based, electrical capacitance-based, and image-based approaches, are quick but not accurate when wood chips come from various sources. Variability in the source material can alter data distributions, undermining the performance of data-driven models. Therefore, there is a need for a robust approach that effectively mitigates the impact of source variability. Previous studies show that manually extracted texture features have the potential to predict wood chip moisture class. Building on this, in this study, we conduct a comprehensive analysis of five distinct texture feature types extracted from wood chip images to predict moisture content. Our findings reveal that a combined feature set incorporating all five texture features achieves an accuracy of 95% and consistently outperforms individual texture features in predicting moisture content. To ensure robust moisture prediction, we propose a domain adaptation method named AdaptMoist that utilizes the texture features to transfer knowledge from one source of wood chip data to another, addressing variability across different domains. We also proposed a criterion for model saving based on adjusted mutual information. The AdaptMoist method improves prediction accuracy across domains by 23%, achieving an average accuracy of 80%, compared to 57% for non-adapted models. These results highlight the effectiveness of AdaptMoist as a robust solution for wood chip moisture content estimation across domains, making it a potential solution for wood chip-reliant industries.
Authors: Xiangyu Mu, Dongliang Zhou, Jie Hou, Haijun Zhang, Weili Guan
Abstract: Mannequin-based clothing displays offer a cost-effective alternative to real-model showcases for online fashion presentation, but lack realism and expressive detail. To overcome this limitation, we introduce a new task called mannequin-to-human (M2H) video generation, which aims to synthesize identity-controllable, photorealistic human videos from footage of mannequins. We propose M2HVideo, a pose-aware and identity-preserving video generation framework that addresses two key challenges: the misalignment between head and body motion, and identity drift caused by temporal modeling. In particular, M2HVideo incorporates a dynamic pose-aware head encoder that fuses facial semantics with body pose to produce consistent identity embeddings across frames. To address the loss of fine facial details due to latent space compression, we introduce a mirror loss applied in pixel space through a denoising diffusion implicit model (DDIM)-based one-step denoising. Additionally, we design a distribution-aware adapter that aligns statistical distributions of identity and clothing features to enhance temporal coherence. Extensive experiments on the UBC fashion dataset, our self-constructed ASOS dataset, and the newly collected MannequinVideos dataset captured on-site demonstrate that M2HVideo achieves superior performance in terms of clothing consistency, identity preservation, and video fidelity in comparison to state-of-the-art methods.
Authors: Haofan Ren, Qingsong Yan, Ming Lu, Rongfeng Lu, Zunjie Zhu
Abstract: Recent advancements in 3D Gaussian Splatting (3DGS) have greatly influenced neural fields, as it enables high-fidelity rendering with impressive visual quality. However, 3DGS has difficulty accurately representing surfaces. In contrast, 2DGS transforms the 3D volume into a collection of 2D planar Gaussian disks. Despite advancements in geometric fidelity, rendering quality remains compromised, highlighting the challenge of achieving both high-quality rendering and precise geometric structures. This indicates that optimizing both geometric and rendering quality in a single training stage is currently unfeasible. To overcome this limitation, we present 2DGS-R, a new method that uses a hierarchical training approach to improve rendering quality while maintaining geometric accuracy. 2DGS-R first trains the original 2D Gaussians with the normal consistency regularization. Then 2DGS-R selects the 2D Gaussians with inadequate rendering quality and applies a novel in-place cloning operation to enhance the 2D Gaussians. Finally, we fine-tune the 2DGS-R model with opacity frozen. Experimental results show that compared to the original 2DGS, our method requires only 1\% more storage and minimal additional training time. Despite this negligible overhead, it achieves high-quality rendering results while preserving fine geometric structures. These findings indicate that our approach effectively balances efficiency with performance, leading to improvements in both visual fidelity and geometric reconstruction accuracy.
Authors: Akhila Kambhatla, Taminul Islam, Khaled R Ahmed
Abstract: The escalating threat of weapon-related violence necessitates automated detection systems capable of pixel-level precision for accurate threat assessment in real-time security applications. Traditional weapon detection approaches rely on object detection frameworks that provide only coarse bounding box localizations, lacking the fine-grained segmentation required for comprehensive threat analysis. Furthermore, existing semantic segmentation models either sacrifice accuracy for computational efficiency or require excessive computational resources incompatible with edge deployment scenarios. This paper presents ArmFormer, a lightweight transformer-based semantic segmentation framework that strategically integrates Convolutional Block Attention Module (CBAM) with MixVisionTransformer architecture to achieve superior accuracy while maintaining computational efficiency suitable for resource-constrained edge devices. Our approach combines CBAM-enhanced encoder backbone with attention-integrated hamburger decoder to enable multi-class weapon segmentation across five categories: handgun, rifle, knife, revolver, and human. Comprehensive experiments demonstrate that ArmFormer achieves state-of-the-art performance with 80.64% mIoU and 89.13% mFscore while maintaining real-time inference at 82.26 FPS. With only 4.886G FLOPs and 3.66M parameters, ArmFormer outperforms heavyweight models requiring up to 48x more computation, establishing it as the optimal solution for deployment on portable security cameras, surveillance drones, and embedded AI accelerators in distributed security infrastructure.
Authors: Shujian Gao, Yuan Wang, Zekuan Yu
Abstract: Semi-supervised medical image segmentation (SSMIS) seeks to match fully supervised performance while sharply reducing annotation cost. Mainstream SSMIS methods rely on \emph{label-space consistency}, yet they overlook the equally critical \emph{representation-space alignment}. Without harmonizing latent features, models struggle to learn representations that are both discriminative and spatially coherent. To this end, we introduce \textbf{Bilateral Alignment in Representation and Label spaces (BARL)}, a unified framework that couples two collaborative branches and enforces alignment in both spaces. For label-space alignment, inspired by co-training and multi-scale decoding, we devise \textbf{Dual-Path Regularization (DPR)} and \textbf{Progressively Cognitive Bias Correction (PCBC)} to impose fine-grained cross-branch consistency while mitigating error accumulation from coarse to fine scales. For representation-space alignment, we conduct region-level and lesion-instance matching between branches, explicitly capturing the fragmented, complex pathological patterns common in medical imagery. Extensive experiments on four public benchmarks and a proprietary CBCT dataset demonstrate that BARL consistently surpasses state-of-the-art SSMIS methods. Ablative studies further validate the contribution of each component. Code will be released soon.
Authors: Yuyang Yu, Zhengwei Chen, Xuemiao Xu, Lei Zhang, Haoxin Yang, Yongwei Nie, Shengfeng He
Abstract: 3D anomaly detection in point-cloud data is critical for industrial quality control, aiming to identify structural defects with high reliability. However, current memory bank-based methods often suffer from inconsistent feature transformations and limited discriminative capacity, particularly in capturing local geometric details and achieving rotation invariance. These limitations become more pronounced when registration fails, leading to unreliable detection results. We argue that point-cloud registration plays an essential role not only in aligning geometric structures but also in guiding feature extraction toward rotation-invariant and locally discriminative representations. To this end, we propose a registration-induced, rotation-invariant feature extraction framework that integrates the objectives of point-cloud registration and memory-based anomaly detection. Our key insight is that both tasks rely on modeling local geometric structures and leveraging feature similarity across samples. By embedding feature extraction into the registration learning process, our framework jointly optimizes alignment and representation learning. This integration enables the network to acquire features that are both robust to rotations and highly effective for anomaly detection. Extensive experiments on the Anomaly-ShapeNet and Real3D-AD datasets demonstrate that our method consistently outperforms existing approaches in effectiveness and generalizability.
Authors: Yudan Ren, Xinlong Wang, Kexin Wang, Tian Xia, Zihan Ma, Zhaowei Li, Xiangrong Bi, Xiao Li, Xiaowei He
Abstract: While brain-inspired artificial intelligence(AI) has demonstrated promising results, current understanding of the parallels between artificial neural networks (ANNs) and human brain processing remains limited: (1) unimodal ANN studies fail to capture the brain's inherent multimodal processing capabilities, and (2) multimodal ANN research primarily focuses on high-level model outputs, neglecting the crucial role of individual neurons. To address these limitations, we propose a novel neuron-level analysis framework that investigates the multimodal information processing mechanisms in vision-language models (VLMs) through the lens of human brain activity. Our approach uniquely combines fine-grained artificial neuron (AN) analysis with fMRI-based voxel encoding to examine two architecturally distinct VLMs: CLIP and METER. Our analysis reveals four key findings: (1) ANs successfully predict biological neurons (BNs) activities across multiple functional networks (including language, vision, attention, and default mode), demonstrating shared representational mechanisms; (2) Both ANs and BNs demonstrate functional redundancy through overlapping neural representations, mirroring the brain's fault-tolerant and collaborative information processing mechanisms; (3) ANs exhibit polarity patterns that parallel the BNs, with oppositely activated BNs showing mirrored activation trends across VLM layers, reflecting the complexity and bidirectional nature of neural information processing; (4) The architectures of CLIP and METER drive distinct BNs: CLIP's independent branches show modality-specific specialization, whereas METER's cross-modal design yields unified cross-modal activation, highlighting the architecture's influence on ANN brain-like properties. These results provide compelling evidence for brain-like hierarchical processing in VLMs at the neuronal level.
Authors: Nusrat Munia, Abdullah Imran
Abstract: Generative models, especially Diffusion Models, have demonstrated remarkable capability in generating high-quality synthetic data, including medical images. However, traditional class-conditioned generative models often struggle to generate images that accurately represent specific medical categories, limiting their usefulness for applications such as skin cancer diagnosis. To address this problem, we propose a classification-induced diffusion model, namely, Class-N-Diff, to simultaneously generate and classify dermoscopic images. Our Class-N-Diff model integrates a classifier within a diffusion model to guide image generation based on its class conditions. Thus, the model has better control over class-conditioned image synthesis, resulting in more realistic and diverse images. Additionally, the classifier demonstrates improved performance, highlighting its effectiveness for downstream diagnostic tasks. This unique integration in our Class-N-Diff makes it a robust tool for enhancing the quality and utility of diffusion model-based synthetic dermoscopic image generation. Our code is available at https://github.com/Munia03/Class-N-Diff.
Authors: Zongjian Li, Zheyuan Liu, Qihui Zhang, Bin Lin, Shenghai Yuan, Zhiyuan Yan, Yang Ye, Wangbo Yu, Yuwei Niu, Li Yuan
Abstract: Instruction-based image editing has achieved remarkable progress; however, models solely trained via supervised fine-tuning often overfit to annotated patterns, hindering their ability to explore and generalize beyond training distributions. To this end, we introduce Edit-R1, a novel post-training framework for instruction-based image editing based on policy optimization. Specifically, we utilize Diffusion Negative-aware Finetuning (DiffusionNFT), a likelihood-free policy optimization method consistent with the flow matching forward process, thereby enabling the use of higher-order samplers and more efficient training. Another key challenge here is the absence of a universal reward model, resulting from the diverse nature of editing instructions and tasks. To bridge this gap, we employ a Multimodal Large Language Model (MLLM) as a unified, training-free reward model, leveraging its output logits to provide fine-grained feedback. Furthermore, we carefully design a low-variance group filtering mechanism to reduce MLLM scoring noise and stabilize optimization. UniWorld-V2, trained with this framework, achieves \textbf{state-of-the-art} results on the ImgEdit and GEdit-Bench benchmarks, scoring 4.49 and 7.83, respectively. Crucially, our framework is model-agnostic, delivering substantial performance gains when applied to diverse base models like Qwen-Image-Edit and FLUX-Kontext, demonstrating its wide applicability. Code and models are publicly available at https://github.com/PKU-YuanGroup/UniWorld-V2.
Authors: Ramon Dalmau, Gabriel Jarry, Philippe Very
Abstract: Aviation's non-CO2 effects, particularly contrails, are a significant contributor to its climate impact. Persistent contrails can evolve into cirrus-like clouds that trap outgoing infrared radiation, with radiative forcing potentially comparable to or exceeding that of aviation's CO2 emissions. While physical models simulate contrail formation, evolution and dissipation, validating and calibrating these models requires linking observed contrails to the flights that generated them, a process known as contrail-to-flight attribution. Satellite-based attribution is challenging due to limited spatial and temporal resolution, as contrails often drift and deform before detection. In this paper, we evaluate an alternative approach using ground-based cameras, which capture contrails shortly after formation at high spatial and temporal resolution, when they remain thin, linear, and visually distinct. Leveraging the ground visible camera contrail sequences (GVCCS) dataset, we introduce a modular framework for attributing contrails observed using ground-based cameras to theoretical contrails derived from aircraft surveillance and meteorological data. The framework accommodates multiple geometric representations and distance metrics, incorporates temporal smoothing, and enables flexible probability-based assignment strategies. This work establishes a strong baseline and provides a modular framework for future research in linking contrails to their source flight.
Authors: Akhila Kambhatla, Ahmed R Khaled
Abstract: Thermal weapon segmentation is crucial for surveillance and security applications, enabling robust detection under lowlight and visually obscured conditions where RGB-based systems fail. While convolutional neural networks (CNNs) dominate thermal segmentation literature, their ability to capture long-range dependencies and fine structural details is limited. Vision Transformers (ViTs), with their global context modeling capabilities, have achieved state-of-the-art results in RGB segmentation tasks, yet their potential in thermal weapon segmentation remains underexplored. This work adapts and evaluates four transformer-based architectures SegFormer, DeepLabV3\+, SegNeXt, and Swin Transformer for binary weapon segmentation on a custom thermal dataset comprising 9,711 images collected from real world surveillance videos and automatically annotated using SAM2. We employ standard augmentation strategies within the MMSegmentation framework to ensure robust model training and fair architectural comparison. Experimental results demonstrate significant improvements in segmentation performance: SegFormer-b5 achieves the highest mIoU (94.15\%) and Pixel Accuracy (97.04\%), while SegFormer-b0 provides the fastest inference speed (98.32 FPS) with competitive mIoU (90.84\%). SegNeXt-mscans offers balanced performance with 85.12 FPS and 92.24\% mIoU, and DeepLabV3\+ R101-D8 reaches 92.76\% mIoU at 29.86 FPS. The transformer architectures demonstrate robust generalization capabilities for weapon detection in low-light and occluded thermal environments, with flexible accuracy-speed trade-offs suitable for diverse real-time security applications.
Authors: Chenxu Li, Zhicai Wang, Yuan Sheng, Xingyu Zhu, Yanbin Hao, Xiang Wang
Abstract: Multimodal Large Language Models (MLLMs) increasingly support dynamic image resolutions. However, current evaluation paradigms primarily assess semantic performance, overlooking the critical question of resolution robustness - whether performance remains stable across varying input resolutions. To address this gap, we introduce \textbf{Res-Bench}, a comprehensive benchmark comprising 14,400 samples across 12 resolution levels and six core capability dimensions. We designed a novel evaluation framework that goes beyond traditional accuracy metrics to capture performance stability. This framework introduces multiple robustness metrics: Spearman's correlation for assessing resolution-performance trends, and Absolute/Relative Continuous Error (ACE/RCE) for measuring performance volatility. Using these metrics, we conducted a large-scale evaluation of leading MLLMs. Our analysis encompasses: (1) model-centric and task-centric robustness examination, (2) investigation of preprocessing strategies including padding and super-resolution, and (3) exploration of fine-tuning for stability enhancement.
Authors: Praveenbalaji Rajendran, Mojtaba Safari, Wenfeng He, Mingzhe Hu, Shansong Wang, Jun Zhou, Xiaofeng Yang
Abstract: Recent advancements in artificial intelligence (AI), particularly foundation models (FMs), have revolutionized medical image analysis, demonstrating strong zero- and few-shot performance across diverse medical imaging tasks, from segmentation to report generation. Unlike traditional task-specific AI models, FMs leverage large corpora of labeled and unlabeled multimodal datasets to learn generalized representations that can be adapted to various downstream clinical applications with minimal fine-tuning. However, despite the rapid proliferation of FM research in medical imaging, the field remains fragmented, lacking a unified synthesis that systematically maps the evolution of architectures, training paradigms, and clinical applications across modalities. To address this gap, this review article provides a comprehensive and structured analysis of FMs in medical image analysis. We systematically categorize studies into vision-only and vision-language FMs based on their architectural foundations, training strategies, and downstream clinical tasks. Additionally, a quantitative meta-analysis of the studies was conducted to characterize temporal trends in dataset utilization and application domains. We also critically discuss persistent challenges, including domain adaptation, efficient fine-tuning, computational constraints, and interpretability along with emerging solutions such as federated learning, knowledge distillation, and advanced prompting. Finally, we identify key future research directions aimed at enhancing the robustness, explainability, and clinical integration of FMs, thereby accelerating their translation into real-world medical practice.
Authors: Yuanzhi Zhu, Eleftherios Tsonis, Lucas Degeorge, Vicky Kalogeiton
Abstract: Diffusion and flow models achieve high generative quality but remain computationally expensive due to slow multi-step sampling. Distillation methods accelerate them by training fast student generators, yet most existing objectives lack a unified theoretical foundation. In this work, we propose Di-Bregman, a compact framework that formulates diffusion distillation as Bregman divergence-based density-ratio matching. This convex-analytic view connects several existing objectives through a common lens. Experiments on CIFAR-10 and text-to-image generation demonstrate that Di-Bregman achieves improved one-step FID over reverse-KL distillation and maintains high visual fidelity compared to the teacher model. Our results highlight Bregman density-ratio matching as a practical and theoretically-grounded route toward efficient one-step diffusion generation.
Authors: Junhao Zhao, Zishuai Liu, Ruili Fang, Jin Lu, Linghan Zhang, Fei Dou
Abstract: The recognition of Activities of Daily Living (ADLs) from event-triggered ambient sensors is an essential task in Ambient Assisted Living, yet existing methods remain constrained by representation-level limitations. Sequence-based approaches preserve temporal order of sensor activations but are sensitive to noise and lack spatial awareness, while image-based approaches capture global patterns and implicit spatial correlations but compress fine-grained temporal dynamics and distort sensor layouts. Naive fusion (e.g., feature concatenation) fail to enforce alignment between sequence- and image-based representation views, underutilizing their complementary strengths. We propose Contrastive Alignment for ADL Recognition from Event-Triggered Sensor Streams (CARE), an end-to-end framework that jointly optimizes representation learning via Sequence-Image Contrastive Alignment (SICA) and classification via cross-entropy, ensuring both cross-representation alignment and task-specific discriminability. CARE integrates (i) time-aware, noise-resilient sequence encoding with (ii) spatially-informed and frequency-sensitive image representations, and employs (iii) a joint contrastive-classification objective for end-to-end learning of aligned and discriminative embeddings. Evaluated on three CASAS datasets, CARE achieves state-of-the-art performance (89.8% on Milan, 88.9% on Cairo, and 73.3% on Kyoto7) and demonstrates robustness to sensor malfunctions and layout variability, highlighting its potential for reliable ADL recognition in smart homes.
Authors: Luca Zanella, Massimiliano Mancini, Yiming Wang, Alessio Tonioni, Elisa Ricci
Abstract: Given a task and a set of steps composing it, Video Step Grounding (VSG) aims to detect which steps are performed in a video. Standard approaches for this task require a labeled training set (e.g., with step-level annotations or narrations), which may be costly to collect. Moreover, they process the full video offline, limiting their applications for scenarios requiring online decisions. Thus, in this work, we explore how to perform VSG online and without training. We achieve this by exploiting the zero-shot capabilities of recent Large Multimodal Models (LMMs). In particular, we use LMMs to predict the step associated with a restricted set of frames, without access to the whole video. We show that this online strategy without task-specific tuning outperforms offline and training-based models. Motivated by this finding, we develop Bayesian Grounding with Large Multimodal Models (BaGLM), further injecting knowledge of past frames into the LMM-based predictions. BaGLM exploits Bayesian filtering principles, modeling step transitions via (i) a dependency matrix extracted through large language models and (ii) an estimation of step progress. Experiments on three datasets show superior performance of BaGLM over state-of-the-art training-based offline methods.
Authors: Ignacio M. De la Jara, Cristian Rodriguez-Opazo, Edison Marrese-Taylor, Felipe Bravo-Marquez
Abstract: Temporal video grounding is a fundamental task in computer vision, aiming to localize a natural language query in a long, untrimmed video. It has a key role in the scientific community, in part due to the large amount of video generated every day. Although we find extensive work in this task, we note that research remains focused on a small selection of video representations, which may lead to architectural overfitting in the long run. To address this issue, we propose an empirical study to investigate the impact of different video features on a classical architecture. We extract features for three well-known benchmarks, Charades-STA, ActivityNet-Captions and YouCookII, using video encoders based on CNNs, temporal reasoning and transformers. Our results show significant differences in the performance of our model by simply changing the video encoder, while also revealing clear patterns and errors derived from the use of certain features, ultimately indicating potential feature complementarity.
Authors: Ani Vanyan, Alvard Barseghyan, Hakob Tamazyan, Tigran Galstyan, Vahan Huroyan, Naira Hovakimyan, Hrant Khachatrian
Abstract: Foundation models have advanced machine learning across various modalities, including images. Recently multiple teams trained foundation models specialized for remote sensing applications. This line of research is motivated by the distinct characteristics of remote sensing imagery, specific applications and types of robustness useful for satellite image analysis. In this work we systematically challenge the idea that specific foundation models are more useful than general-purpose vision foundation models, at least in the small scale. First, we design a simple benchmark that measures generalization of remote sensing models towards images with lower resolution for two downstream tasks. Second, we train iBOT, a self-supervised vision encoder, on MillionAID, an ImageNet-scale satellite imagery dataset, with several modifications specific to remote sensing. We show that none of those pretrained models bring consistent improvements upon general-purpose baselines at the ViT-B scale.
Authors: Shraman Pramanick, Effrosyni Mavroudi, Yale Song, Rama Chellappa, Lorenzo Torresani, Triantafyllos Afouras
Abstract: We introduce ED-VTG, a method for fine-grained video temporal grounding utilizing multi-modal large language models. Our approach harnesses the capabilities of multimodal LLMs to jointly process text and video, in order to effectively localize natural language queries in videos through a two-stage process. Rather than being directly grounded, language queries are initially transformed into enriched sentences that incorporate missing details and cues to aid in grounding. In the second stage, these enriched queries are grounded, using a lightweight decoder, which specializes at predicting accurate boundaries conditioned on contextualized representations of the enriched queries. To mitigate noise and reduce the impact of hallucinations, our model is trained with a multiple-instance-learning objective that dynamically selects the optimal version of the query for each training sample. We demonstrate state-of-the-art results across various benchmarks in temporal video grounding and paragraph grounding settings. Experiments reveal that our method significantly outperforms all previously proposed LLM-based temporal grounding approaches and is either superior or comparable to specialized models, while maintaining a clear advantage against them in zero-shot evaluation scenarios.
Authors: Yutong Zhong
Abstract: Multimodal 3D grounding has garnered considerable interest in Vision-Language Models (VLMs) \cite{yin2025spatial} for advancing spatial reasoning in complex environments. However, these models suffer from a severe "2D semantic bias" that arises from over-reliance on 2D image features for coarse localization, largely disregarding 3D geometric inputs and resulting in suboptimal fusion performance. In this paper, we propose a novel training framework called What-Where Representation Re-Forming (W2R2) to tackle this issue via disentangled representation learning and targeted shortcut suppression. Our approach fundamentally reshapes the model's internal space by designating 2D features as semantic beacons for "What" identification and 3D features as spatial anchors for "Where" localization, enabling precise 3D grounding without modifying inference architecture. Key components include a dual-objective loss function with an Alignment Loss that supervises fused predictions using adapted cross-entropy for multimodal synergy, and a Pseudo-Label Loss that penalizes overly effective 2D-dominant pseudo-outputs via a margin-based mechanism. Experiments conducted on ScanRefer and ScanQA demonstrate the effectiveness of W2R2, with significant gains in localization accuracy and robustness, particularly in cluttered outdoor scenes.
Authors: Syed Konain Abbas, Sandip Purnapatra, M. G. Sarwar Murshed, Conor Miller-Lynch, Lambert Igene, Soumyabrata Dey, Stephanie Schuckers, Faraz Hussain
Abstract: Large fingerprint datasets, while important for training and evaluation, are time-consuming and expensive to collect and require strict privacy measures. Researchers are exploring the use of synthetic fingerprint data to address these issues. This paper presents a novel approach for generating synthetic fingerprint images (both spoof and live), addressing concerns related to privacy, cost, and accessibility in biometric data collection. Our approach utilizes conditional StyleGAN2-ADA and StyleGAN3 architectures to produce high-resolution synthetic live fingerprints, conditioned on specific finger identities (thumb through little finger). Additionally, we employ CycleGANs to translate these into realistic spoof fingerprints, simulating a variety of presentation attack materials (e.g., EcoFlex, Play-Doh). These synthetic spoof fingerprints are crucial for developing robust spoof detection systems. Through these generative models, we created two synthetic datasets (DB2 and DB3), each containing 1,500 fingerprint images of all ten fingers with multiple impressions per finger, and including corresponding spoofs in eight material types. The results indicate robust performance: our StyleGAN3 model achieves a Fr\'echet Inception Distance (FID) as low as 5, and the generated fingerprints achieve a True Accept Rate of 99.47% at a 0.01% False Accept Rate. The StyleGAN2-ADA model achieved a TAR of 98.67% at the same 0.01% FAR. We assess fingerprint quality using standard metrics (NFIQ2, MINDTCT), and notably, matching experiments confirm strong privacy preservation, with no significant evidence of identity leakage, confirming the strong privacy-preserving properties of our synthetic datasets.
Authors: Mohammad R. Salmanpour, Sonya Falahati, Amir Hossein Pouria, Amin Mousavi, Somayeh Sadat Mehrnia, Morteza Alizadeh, Arman Gorji, Zeinab Farsangi, Alireza Safarian, Mehdi Maghsudi, Carlos Uribe, Arman Rahmim, Ren Yuan
Abstract: Lung cancer remains the leading cause of cancer mortality, with CT imaging central to screening, prognosis, and treatment. Manual segmentation is variable and time-intensive, while deep learning (DL) offers automation but faces barriers to clinical adoption. Guided by the Knowledge-to-Action framework, this study develops a clinician-in-the-loop DL pipeline to enhance reproducibility, prognostic accuracy, and clinical trust. Multi-center CT data from 999 patients across 12 public datasets were analyzed using five DL models (3D Attention U-Net, ResUNet, VNet, ReconNet, SAM-Med3D), benchmarked against expert contours on whole and click-point cropped images. Segmentation reproducibility was assessed using 497 PySERA-extracted radiomic features via Spearman correlation, ICC, Wilcoxon tests, and MANOVA, while prognostic modeling compared supervised (SL) and semi-supervised learning (SSL) across 38 dimensionality reduction strategies and 24 classifiers. Six physicians qualitatively evaluated masks across seven domains, including clinical meaningfulness, boundary quality, prognostic value, trust, and workflow integration. VNet achieved the best performance (Dice = 0.83, IoU = 0.71), radiomic stability (mean correlation = 0.76, ICC = 0.65), and predictive accuracy under SSL (accuracy = 0.88, F1 = 0.83). SSL consistently outperformed SL across models. Radiologists favored VNet for peritumoral representation and smoother boundaries, preferring AI-generated initial masks for refinement rather than replacement. These results demonstrate that integrating VNet with SSL yields accurate, reproducible, and clinically trusted CT-based lung cancer prognosis, highlighting a feasible path toward physician-centered AI translation.
Authors: Md Ahmed Al Muzaddid, William J. Beksi
Abstract: Advanced feature extraction methods have significantly contributed to enhancing the task of person re-identification. In addition, modifications to objective functions have been developed to further improve performance. Nonetheless, selecting better class representatives is an underexplored area of research that can also lead to advancements in re-identification performance. Although past works have experimented with using the centroid of a gallery image class during training, only a few have investigated alternative representations during the retrieval stage. In this paper, we demonstrate that these prior techniques yield suboptimal results in terms of re-identification metrics. To address the re-identification problem, we propose a generalized selection method that involves choosing representations that are not limited to class centroids. Our approach strikes a balance between accuracy and mean average precision, leading to improvements beyond the state of the art. For example, the actual number of representations per class can be adjusted to meet specific application requirements. We apply our methodology on top of multiple re-identification embeddings, and in all cases it substantially improves upon contemporary results
Authors: Deepak Sridhar, Kartikeya Bhardwaj, Jeya Pradha Jeyaraj, Nuno Vasconcelos, Ankita Nayak, Harris Teague
Abstract: Video reasoning using Large Multimodal Models (LMMs) relies on costly reinforcement learning (RL) and verbose chain-of-thought, resulting in substantial computational overhead during both training and inference. Moreover, the mechanisms that control the thinking process in these reasoning models are very limited. In this paper, using entropy of the model's output as a signal, we discover that the high-quality models go through a series of micro-explorations and micro-exploitations which keep the reasoning process grounded (i.e., avoid excessive randomness while the model is exploring or thinking through an answer). We further observe that once this "thinking" process is over, more accurate models demonstrate a better convergence by reducing the entropy significantly via a final exploitation phase (i.e., a more certain convergence towards a solution trajectory). We then use these novel, theoretically-grounded insights to tune the model's behavior directly at inference, without using any RL or supervised fine-tuning. Specifically, during inference, our proposed approach called V-Reason (Video-Reason) adapts the value cache of the LMM via a few optimization steps on a small, trainable controller using an entropy-based objective, i.e., no supervision from any dataset or RL is necessary. This tuning improves the model's micro-exploration and exploitation behavior during inference. Our experiments show that our proposed method achieves significant improvements over the base instruction-tuned models across several video reasoning datasets, narrowing the gap with RL-trained models to within 0.6% average accuracy without any training, while offering massive efficiency benefits: output tokens are reduced by 58.6% compared to the RL model.
Authors: Masoud Khairi Atani, Alon Harell, Hyomin Choi, Runyu Yang, Fabien Racape, Ivan V. Bajic
Abstract: The trade-off between general-purpose foundation vision models and their specialized counterparts is critical for efficient feature coding design and is not yet fully understood. We investigate this trade-off by comparing the feature versatility of the general-purpose Hiera encoder against the segmentation-specialized Segment Anything Model 2 (SAM2). Using a lightweight, trainable neck to probe the adaptability of their frozen features, we quantify the information-theoretic cost of specialization. Our results reveal that while SAM2's specialization is highly effective for spatially-related tasks like depth estimation, it comes at a cost. The specialized SAM2 encoder underperforms its generalist predecessor, Hiera, on conceptually distant tasks such as pose estimation and image captioning, demonstrating a measurable loss of broader semantic information. A novel cross-neck analysis on SAM2 reveals that each level of adaptation creates a further representational bottleneck. Our analysis illuminates these trade-offs in feature universality, providing a quantitative foundation for designing efficient feature coding and adaptation strategies for diverse downstream applications.
Authors: Zhe Luo, Wenjing Jia, Stuart Perry
Abstract: Three-dimensional (3D) point clouds are becoming increasingly vital in applications such as autonomous driving, augmented reality, and immersive communication, demanding real-time processing and low latency. However, their large data volumes and bandwidth constraints hinder the deployment of high-quality services in resource-limited environments. Progres- sive coding, which allows for decoding at varying levels of detail, provides an alternative by allowing initial partial decoding with subsequent refinement. Although recent learning-based point cloud geometry coding methods have achieved notable success, their fixed latent representation does not support progressive decoding. To bridge this gap, we propose ProDAT, a novel density-aware tail-drop mechanism for progressive point cloud coding. By leveraging density information as a guidance signal, latent features and coordinates are decoded adaptively based on their significance, therefore achieving progressive decoding at multiple bitrates using one single model. Experimental results on benchmark datasets show that the proposed ProDAT not only enables progressive coding but also achieves superior coding efficiency compared to state-of-the-art learning-based coding techniques, with over 28.6% BD-rate improvement for PSNR- D2 on SemanticKITTI and over 18.15% for ShapeNet
Authors: Jad Berjawi, Yoann Dupas, Christophe C'erin
Abstract: Multimodal object detection improves robustness in chal- lenging conditions by leveraging complementary cues from multiple sensor modalities. We introduce Filtered Multi- Modal Cross Attention Fusion (FMCAF), a preprocess- ing architecture designed to enhance the fusion of RGB and infrared (IR) inputs. FMCAF combines a frequency- domain filtering block (Freq-Filter) to suppress redun- dant spectral features with a cross-attention-based fusion module (MCAF) to improve intermodal feature sharing. Unlike approaches tailored to specific datasets, FMCAF aims for generalizability, improving performance across different multimodal challenges without requiring dataset- specific tuning. On LLVIP (low-light pedestrian detec- tion) and VEDAI (aerial vehicle detection), FMCAF outper- forms traditional fusion (concatenation), achieving +13.9% mAP@50 on VEDAI and +1.1% on LLVIP. These results support the potential of FMCAF as a flexible foundation for robust multimodal fusion in future detection pipelines.
Authors: Ruitong Gan, Junran Peng, Yang Liu, Chuanchen Luo, Qing Li, Zhaoxiang Zhang
Abstract: Planes are fundamental primitives of 3D sences, especially in man-made environments such as indoor spaces and urban streets. Representing these planes in a structured and parameterized format facilitates scene editing and physical simulations in downstream applications. Recently, Gaussian Splatting (GS) has demonstrated remarkable effectiveness in the Novel View Synthesis task, with extensions showing great potential in accurate surface reconstruction. However, even state-of-the-art GS representations often struggle to reconstruct planar regions with sufficient smoothness and precision. To address this issue, we propose GSPlane, which recovers accurate geometry and produces clean and well-structured mesh connectivity for plane regions in the reconstructed scene. By leveraging off-the-shelf segmentation and normal prediction models, GSPlane extracts robust planar priors to establish structured representations for planar Gaussian coordinates, which help guide the training process by enforcing geometric consistency. To further enhance training robustness, a Dynamic Gaussian Re-classifier is introduced to adaptively reclassify planar Gaussians with persistently high gradients as non-planar, ensuring more reliable optimization. Furthermore, we utilize the optimized planar priors to refine the mesh layouts, significantly improving topological structure while reducing the number of vertices and faces. We also explore applications of the structured planar representation, which enable decoupling and flexible manipulation of objects on supportive planes. Extensive experiments demonstrate that, with no sacrifice in rendering quality, the introduction of planar priors significantly improves the geometric accuracy of the extracted meshes across various baselines.
Authors: Xiaogang Xu, Jian Wang, Yunfan Lu, Ruihang Chu, Ruixing Wang, Jiafei Wu, Bei Yu, Liang Lin
Abstract: Diffusion-based methods, leveraging pre-trained large models like Stable Diffusion via ControlNet, have achieved remarkable performance in several low-level vision tasks. However, Pre-Trained Diffusion-Based (PTDB) methods often sacrifice content fidelity to attain higher perceptual realism. This issue is exacerbated in low-light scenarios, where severely degraded information caused by the darkness limits effective control. We identify two primary causes of fidelity loss: the absence of suitable conditional latent modeling and the lack of bidirectional interaction between the conditional latent and noisy latent in the diffusion process. To address this, we propose a novel optimization strategy for conditioning in pre-trained diffusion models, enhancing fidelity while preserving realism and aesthetics. Our method introduces a mechanism to recover spatial details lost during VAE encoding, i.e., a latent refinement pipeline incorporating generative priors. Additionally, the refined latent condition interacts dynamically with the noisy latent, leading to improved restoration performance. Our approach is plug-and-play, seamlessly integrating into existing diffusion networks to provide more effective control. Extensive experiments demonstrate significant fidelity improvements in PTDB methods.
Authors: Hodaka Kawachi, Tomoya Nakamura, Hiroaki Santo, SaiKiran Kumar Tedla, Trevor Dalton Canham, Yasushi Yagi, Michael S. Brown
Abstract: This paper introduces a method for using LED-based environmental lighting to produce visually imperceptible watermarks for consumer cameras. Our approach optimizes an LED light source's spectral profile to be minimally visible to the human eye while remaining highly detectable by typical consumer cameras. The method jointly considers the human visual system's sensitivity to visible spectra, modern consumer camera sensors' spectral sensitivity, and narrowband LEDs' ability to generate broadband spectra perceived as "white light" (specifically, D65 illumination). To ensure imperceptibility, we employ spectral modulation rather than intensity modulation. Unlike conventional visible light communication, our approach enables watermark extraction at standard low frame rates (30-60 fps). While the information transfer rate is modest-embedding 128 bits within a 10-second video clip-this capacity is sufficient for essential metadata supporting privacy protection and content verification.
Authors: Xin Gao, Jiyao Liu, Guanghao Li, Yueming Lyu, Jianxiong Gao, Weichen Yu, Ningsheng Xu, Liang Wang, Caifeng Shan, Ziwei Liu, Chenyang Si
Abstract: Recent advancements have explored text-to-image diffusion models for synthesizing out-of-distribution (OOD) samples, substantially enhancing the performance of OOD detection. However, existing approaches typically rely on perturbing text-conditioned embeddings, resulting in semantic instability and insufficient shift diversity, which limit generalization to realistic OOD. To address these challenges, we propose GOOD, a novel and flexible framework that directly guides diffusion sampling trajectories towards OOD regions using off-the-shelf in-distribution (ID) classifiers. GOOD incorporates dual-level guidance: (1) Image-level guidance based on the gradient of log partition to reduce input likelihood, drives samples toward low-density regions in pixel space. (2) Feature-level guidance, derived from k-NN distance in the classifier's latent space, promotes sampling in feature-sparse regions. Hence, this dual-guidance design enables more controllable and diverse OOD sample generation. Additionally, we introduce a unified OOD score that adaptively combines image and feature discrepancies, enhancing detection robustness. We perform thorough quantitative and qualitative analyses to evaluate the effectiveness of GOOD, demonstrating that training with samples generated by GOOD can notably enhance OOD detection performance.
Authors: WenBo Xu, Liu Liu, Li Zhang, Ran Zhang, Hao Wu, Dan Guo, Meng Wang
Abstract: Articulated objects, such as laptops and drawers, exhibit significant challenges for 3D reconstruction and pose estimation due to their multi-part geometries and variable joint configurations, which introduce structural diversity across different states. To address these challenges, we propose KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation, a unified framework for reconstructing diverse articulated instances and pose estimation from single view input. Specifically, we first encode complete geometry (SDFs), joint angles, and part segmentation into a structured latent space via a novel Kinematic-Aware VAE (KA-VAE). In addition, we employ two conditional diffusion models: one for regressing global pose (SE(3)) and joint parameters, and another for generating the kinematic-aware latent code from partial observations. Finally, we produce an iterative optimization module that bidirectionally refines reconstruction accuracy and kinematic parameters via Chamfer-distance minimization while preserving articulation constraints. Experimental results on synthetic, semi-synthetic, and real-world datasets demonstrate the effectiveness of our approach in accurately reconstructing articulated objects and estimating their kinematic properties.
Authors: Yinghui Wang, Xinyu Zhang, Peng Du
Abstract: Generating editable, parametric CAD models from a single image holds great potential to lower the barriers of industrial concept design. However, current multi-modal large language models (MLLMs) still struggle with accurately inferring 3D geometry from 2D images due to limited spatial reasoning capabilities. We address this limitation by introducing GACO-CAD, a novel two-stage post-training framework. It is designed to achieve a joint objective: simultaneously improving the geometric accuracy of the generated CAD models and encouraging the use of more concise modeling procedures. First, during supervised fine-tuning, we leverage depth and surface normal maps as dense geometric priors, combining them with the RGB image to form a multi-channel input. In the context of single-view reconstruction, these priors provide complementary spatial cues that help the MLLM more reliably recover 3D geometry from 2D observations. Second, during reinforcement learning, we introduce a group length reward that, while preserving high geometric fidelity, promotes the generation of more compact and less redundant parametric modeling sequences. A simple dynamic weighting strategy is adopted to stabilize training. Experiments on the DeepCAD and Fusion360 datasets show that GACO-CAD achieves state-of-the-art performance under the same MLLM backbone, consistently outperforming existing methods in terms of code validity, geometric accuracy, and modeling conciseness.
Authors: Roland Croft, Brian Du, Darcy Joseph, Sharath Kumar
Abstract: Face Recognition (FR) models have been shown to be vulnerable to adversarial examples that subtly alter benign facial images, exposing blind spots in these systems, as well as protecting user privacy. End-to-end FR systems first obtain preprocessed faces from diverse facial imagery prior to computing the similarity of the deep feature embeddings. Whilst face preprocessing is a critical component of FR systems, and hence adversarial attacks against them, we observe that this preprocessing is often overlooked in blackbox settings. Our study seeks to investigate the transferability of several out-of-the-box state-of-the-art adversarial attacks against FR when applied against different preprocessing techniques used in a blackbox setting. We observe that the choice of face detection model can degrade the attack success rate by up to 78%, whereas choice of interpolation method during downsampling has relatively minimal impacts. Furthermore, we find that the requirement for facial preprocessing even degrades attack strength in a whitebox setting, due to the unintended interaction of produced noise vectors against face detection models. Based on these findings, we propose a preprocessing-invariant method using input transformations that improves the transferability of the studied attacks by up to 27%. Our findings highlight the importance of preprocessing in FR systems, and the need for its consideration towards improving the adversarial generalisation of facial adversarial examples.
Authors: Feihong Yan, Peiru Wang, Yao Zhu, Kaiyu Pang, Qingyan Wei, Huiqi Li, Linfeng Zhang
Abstract: Masked Autoregressive (MAR) models promise better efficiency in visual generation than autoregressive (AR) models for the ability of parallel generation, yet their acceleration potential remains constrained by the modeling complexity of spatially correlated visual tokens in a single step. To address this limitation, we introduce Generation then Reconstruction (GtR), a training-free hierarchical sampling strategy that decomposes generation into two stages: structure generation establishing global semantic scaffolding, followed by detail reconstruction efficiently completing remaining tokens. Assuming that it is more difficult to create an image from scratch than to complement images based on a basic image framework, GtR is designed to achieve acceleration by computing the reconstruction stage quickly while maintaining the generation quality by computing the generation stage slowly. Moreover, observing that tokens on the details of an image often carry more semantic information than tokens in the salient regions, we further propose Frequency-Weighted Token Selection (FTS) to offer more computation budget to tokens on image details, which are localized based on the energy of high frequency information. Extensive experiments on ImageNet class-conditional and text-to-image generation demonstrate 3.72x speedup on MAR-H while maintaining comparable quality (e.g., FID: 1.59, IS: 304.4 vs. original 1.59, 299.1), substantially outperforming existing acceleration methods across various model scales and generation tasks. Our codes will be released in https://github.com/feihongyan1/GtR.
Authors: Yingzi Han, Jiakai He, Chuanlong Xie, Jianping Li
Abstract: Automated plankton recognition models face significant challenges during real-world deployment due to distribution shifts (Out-of-Distribution, OoD) between training and test data. This stems from plankton's complex morphologies, vast species diversity, and the continuous discovery of novel species, which leads to unpredictable errors during inference. Despite rapid advancements in OoD detection methods in recent years, the field of plankton recognition still lacks a systematic integration of the latest computer vision developments and a unified benchmark for large-scale evaluation. To address this, this paper meticulously designed a series of OoD benchmarks simulating various distribution shift scenarios based on the DYB-PlanktonNet dataset \cite{875n-f104-21}, and systematically evaluated twenty-two OoD detection methods. Extensive experimental results demonstrate that the ViM \cite{wang2022vim} method significantly outperforms other approaches in our constructed benchmarks, particularly excelling in Far-OoD scenarios with substantial improvements in key metrics. This comprehensive evaluation not only provides a reliable reference for algorithm selection in automated plankton recognition but also lays a solid foundation for future research in plankton OoD detection. To our knowledge, this study marks the first large-scale, systematic evaluation and analysis of Out-of-Distribution data detection methods in plankton recognition. Code is available at https://github.com/BlackJack0083/PlanktonOoD.
Authors: Haonan He, Yufeng Zheng, Jie Song
Abstract: Photorealistic 3D head avatars are vital for telepresence, gaming, and VR. However, most methods focus solely on facial regions, ignoring natural hand-face interactions, such as a hand resting on the chin or fingers gently touching the cheek, which convey cognitive states like pondering. In this work, we present a novel framework that jointly learns detailed head avatars and the non-rigid deformations induced by hand-face interactions. There are two principal challenges in this task. First, naively tracking hand and face separately fails to capture their relative poses. To overcome this, we propose to combine depth order loss with contact regularization during pose tracking, ensuring correct spatial relationships between the face and hand. Second, no publicly available priors exist for hand-induced deformations, making them non-trivial to learn from monocular videos. To address this, we learn a PCA basis specific to hand-induced facial deformations from a face-hand interaction dataset. This reduces the problem to estimating a compact set of PCA parameters rather than a full spatial deformation field. Furthermore, inspired by physics-based simulation, we incorporate a contact loss that provides additional supervision, significantly reducing interpenetration artifacts and enhancing the physical plausibility of the results. We evaluate our approach on RGB(D) videos captured by an iPhone. Additionally, to better evaluate the reconstructed geometry, we construct a synthetic dataset of avatars with various types of hand interactions. We show that our method can capture better appearance and more accurate deforming geometry of the face than SOTA surface reconstruction methods.
Authors: Vaibhav Rathore, Divyam Gupta, Biplab Banerjee
Abstract: Generalized Category Discovery (GCD) aims to classify test-time samples into either seen categories** -- available during training -- or novel ones, without relying on label supervision. Most existing GCD methods assume simultaneous access to labeled and unlabeled data during training and arising from the same domain, limiting applicability in open-world scenarios involving distribution shifts. Domain Generalization with GCD (DG-GCD) lifts this constraint by requiring models to generalize to unseen domains containing novel categories, without accessing targetdomain data during training. The only prior DG-GCD method, DG2CD-Net, relies on episodic training with multiple synthetic domains and task vector aggregation, incurring high computational cost and error accumulation. We propose HIDISC, a hyperbolic representation learning framework that achieves domain and category-level generalization without episodic simulation. To expose the model to minimal but diverse domain variations, we augment the source domain using GPT-guided diffusion, avoiding overfitting while maintaining efficiency. To structure the representation space, we introduce Tangent CutMix, a curvature-aware interpolation that synthesizes pseudo-novel samples in tangent space, preserving manifold consistency. A unified loss -- combining penalized Busemann alignment, hybrid hyperbolic contrastive regularization, and adaptive outlier repulsion -- **facilitates compact, semantically structured embeddings. A learnable curvature parameter further adapts the geometry to dataset complexity. HIDISC achieves state-of-the-art results on PACS , Office-Home , and DomainNet, consistently outperforming the existing Euclidean and hyperbolic (DG)-GCD baselines.
Authors: Pu Zhang, Yuwei Li, Xingyuan Xian, Guoming Tang
Abstract: As the capabilities of Vision-Language Models (VLMs) advance, they can process increasingly large inputs, which, unlike in LLMs, generates significant visual token redundancy and leads to prohibitive inference costs. While many methods aim to reduce these costs by pruning visual tokens, existing approaches, whether based on attention or diversity, typically neglect the guidance of the text prompt and thus fail to prioritize task relevance. In this work, we propose a novel, zero-shot method that reframes the problem by introducing a prompt-aware perspective, explicitly modeling visual token pruning as a balance between task relevance and information diversity. Our hierarchical approach first selects a core set of task-relevant visual tokens and then supplements them with diversity tokens to preserve broader context. Experiments across multiple models and benchmarks show that our method achieves performance that matches or surpasses the state-of-the-art with only minimal accuracy loss, even when pruning up to 90\% of the tokens. Furthermore, these gains are accompanied by significant reductions in GPU memory footprint and inference latency.
Authors: M Saifuzzaman Rafat, Mohd Ruhul Ameen, Akif Islam, Abu Saleh Musa Miah, Jungpil Shin
Abstract: The great rivers of Bangladesh, arteries of commerce and sustenance, are also agents of relentless destruction. Each year, they swallow whole villages and vast tracts of farmland, erasing communities from the map and displacing thousands of families. To track this slow-motion catastrophe has, until now, been a Herculean task for human analysts. Here we show how a powerful general-purpose vision model, the Segment Anything Model (SAM), can be adapted to this task with remarkable precision. To do this, we assembled a new dataset - a digital chronicle of loss compiled from historical Google Earth imagery of Bangladesh's most vulnerable regions, including Mokterer Char Union, Kedarpur Union, Balchipara village, and Chowhali Upazila, from 2003 to 2025. Crucially, this dataset is the first to include manually annotated data on the settlements that have vanished beneath the water. Our method first uses a simple color-channel analysis to provide a rough segmentation of land and water, and then fine-tunes SAM's mask decoder to recognize the subtle signatures of riverbank erosion. The resulting model demonstrates a keen eye for this destructive process, achieving a mean Intersection over Union of 86.30% and a Dice score of 92.60% - a performance that significantly surpasses traditional methods and off-the-shelf deep learning models. This work delivers three key contributions: the first annotated dataset of disappeared settlements in Bangladesh due to river erosion; a specialized AI model fine-tuned for this critical task; and a method for quantifying land loss with compelling visual evidence. Together, these tools provide a powerful new lens through which policymakers and disaster management agencies can monitor erosion, anticipate its trajectory, and ultimately protect the vulnerable communities in its path.
Authors: Nirai Hayakawa, Kazumasa Shimari, Kazuma Yamasaki, Hirotatsu Hoshikawa, Rikuto Tsuchida, Kenichi Matsumoto
Abstract: Recently, research on predicting match outcomes in esports has been actively conducted, but much of it is based on match log data and statistical information. This research targets the FPS game VALORANT, which requires complex strategies, and aims to build a round outcome prediction model by analyzing minimap information in match footage. Specifically, based on the video recognition model TimeSformer, we attempt to improve prediction accuracy by incorporating detailed tactical features extracted from minimap information, such as character position information and other in-game events. This paper reports preliminary results showing that a model trained on a dataset augmented with such tactical event labels achieved approximately 81% prediction accuracy, especially from the middle phases of a round onward, significantly outperforming a model trained on a dataset with the minimap information itself. This suggests that leveraging tactical features from match footage is highly effective for predicting round outcomes in VALORANT.
Authors: Bingrong Liu, Jun Shi, Yushan Zheng
Abstract: Class-incremental learning (CIL) for endoscopic image analysis is crucial for real-world clinical applications, where diagnostic models should continuously adapt to evolving clinical data while retaining performance on previously learned ones. However, existing replay-based CIL methods fail to effectively mitigate catastrophic forgetting due to severe domain discrepancies and class imbalance inherent in endoscopic imaging. To tackle these challenges, we propose EndoCIL, a novel and unified CIL framework specifically tailored for endoscopic image diagnosis. EndoCIL incorporates three key components: Maximum Mean Discrepancy Based Replay (MDBR), employing a distribution-aligned greedy strategy to select diverse and representative exemplars, Prior Regularized Class Balanced Loss (PRCBL), designed to alleviate both inter-phase and intra-phase class imbalance by integrating prior class distributions and balance weights into the loss function, and Calibration of Fully-Connected Gradients (CFG), which adjusts the classifier gradients to mitigate bias toward new classes. Extensive experiments conducted on four public endoscopic datasets demonstrate that EndoCIL generally outperforms state-of-the-art CIL methods across varying buffer sizes and evaluation metrics. The proposed framework effectively balances stability and plasticity in lifelong endoscopic diagnosis, showing promising potential for clinical scalability and deployment.
Authors: Mika Feng, Pierre Gallin-Martel, Koichi Ito, Takafumi Aoki
Abstract: Face recognition systems are designed to be robust against variations in head pose, illumination, and image blur during capture. However, malicious actors can exploit these systems by presenting a face photo of a registered user, potentially bypassing the authentication process. Such spoofing attacks must be detected prior to face recognition. In this paper, we propose a DINOv2-based spoofing attack detection method to discern minute differences between live and spoofed face images. Specifically, we employ DINOv2 with registers to extract generalizable features and to suppress perturbations in the attention mechanism, which enables focused attention on essential and minute features. We demonstrate the effectiveness of the proposed method through experiments conducted on the dataset provided by ``The 6th Face Anti-Spoofing Workshop: Unified Physical-Digital Attacks Detection@ICCV2025'' and SiW dataset.
Authors: Yingqi Fan, Anhao Zhao, Jinlan Fu, Junlong Tong, Hui Su, Yijie Pan, Wei Zhang, Xiaoyu Shen
Abstract: Multimodal Large Language Models (MLLMs) have achieved strong performance across vision-language tasks, but suffer from significant computational overhead due to the quadratic growth of attention computations with the number of multimodal tokens. Though efforts have been made to prune tokens in MLLMs, \textit{they lack a fundamental understanding of how MLLMs process and fuse multimodal information.} Through systematic analysis, we uncover a \textbf{three-stage} cross-modal interaction process: (1) Shallow layers recognize task intent, with visual tokens acting as passive attention sinks; (2) Cross-modal fusion occurs abruptly in middle layers, driven by a few critical visual tokens; (3) Deep layers discard vision tokens, focusing solely on linguistic refinement. Based on these findings, we propose \emph{VisiPruner}, a training-free pruning framework that reduces up to 99\% of vision-related attention computations and 53.9\% of FLOPs on LLaVA-v1.5 7B. It significantly outperforms existing token pruning methods and generalizes across diverse MLLMs. Beyond pruning, our insights further provide actionable guidelines for training efficient MLLMs by aligning model architecture with its intrinsic layer-wise processing dynamics. Our code is available at: https://github.com/EIT-NLP/VisiPruner.
Authors: Zhuo Cao, Heming Du, Bingqing Zhang, Xin Yu, Xue Li, Sen Wang
Abstract: Existing Moment retrieval (MR) methods focus on Single-Moment Retrieval (SMR). However, one query can correspond to multiple relevant moments in real-world applications. This makes the existing datasets and methods insufficient for video temporal grounding. By revisiting the gap between current MR tasks and real-world applications, we introduce a high-quality datasets called QVHighlights Multi-Moment Dataset (QV-M$^2$), along with new evaluation metrics tailored for multi-moment retrieval (MMR). QV-M$^2$ consists of 2,212 annotations covering 6,384 video segments. Building on existing efforts in MMR, we propose a framework called FlashMMR. Specifically, we propose a Multi-moment Post-verification module to refine the moment boundaries. We introduce constrained temporal adjustment and subsequently leverage a verification module to re-evaluate the candidate segments. Through this sophisticated filtering pipeline, low-confidence proposals are pruned, and robust multi-moment alignment is achieved. We retrain and evaluate 6 existing MR methods on QV-M$^2$ and QVHighlights under both SMR and MMR settings. Results show that QV-M$^2$ serves as an effective benchmark for training and evaluating MMR models, while FlashMMR provides a strong baseline. Specifically, on QV-M$^2$, it achieves improvements over prior SOTA method by 3.00% on G-mAP, 2.70% on mAP@3+tgt, and 2.56% on mR@3. The proposed benchmark and method establish a foundation for advancing research in more realistic and challenging video temporal grounding scenarios. Code is released at https://github.com/Zhuo-Cao/QV-M2.
Authors: Akihito Yoshii, Ryosuke Sonoda, Ramya Srinivasan
Abstract: Existing deepfake detection methods often exhibit bias, lack transparency, and fail to capture temporal information, leading to biased decisions and unreliable results across different demographic groups. In this paper, we propose a fairness-aware deepfake detection framework that integrates temporal feature learning and demographic-aware data augmentation to enhance fairness and interpretability. Our method leverages sequence-based clustering for temporal modeling of deepfake videos and concept extraction to improve detection reliability while also facilitating interpretable decisions for non-expert users. Additionally, we introduce a demography-aware data augmentation method that balances underrepresented groups and applies frequency-domain transformations to preserve deepfake artifacts, thereby mitigating bias and improving generalization. Extensive experiments on FaceForensics++, DFD, Celeb-DF, and DFDC datasets using state-of-the-art (SoTA) architectures (Xception, ResNet) demonstrate the efficacy of the proposed method in obtaining the best tradeoff between fairness and accuracy when compared to SoTA.
Authors: Luis Wiedmann, Orr Zohar, Amir Mahla, Xiaohan Wang, Rui Li, Thibaud Frere, Leandro von Werra, Aritra Roy Gosthipaty, Andr\'es Marafioti
Abstract: The advancement of vision-language models (VLMs) is hampered by a fragmented landscape of inconsistent and contaminated public datasets. We introduce FineVision, a meticulously collected, curated, and unified corpus of 24 million samples - the largest open resource of its kind. We unify more than 200 sources into 185 subsets via a semi-automated, human-in-the-loop pipeline: automation performs bulk ingestion and schema mapping, while reviewers audit mappings and spot-check outputs to verify faithful consumption of annotations, appropriate formatting and diversity, and safety; issues trigger targeted fixes and re-runs. The workflow further applies rigorous de-duplication within and across sources and decontamination against 66 public benchmarks. FineVision also encompasses agentic/GUI tasks with a unified action space; reviewers validate schemas and inspect a sample of trajectories to confirm executable fidelity. Models trained on FineVision consistently outperform those trained on existing open mixtures across a broad evaluation suite, underscoring the benefits of scale, data hygiene, and balanced automation with human oversight. We release the corpus and curation tools to accelerate data-centric VLM research.
Authors: Katie Luo, Jingwei Ji, Tong He, Runsheng Xu, Yichen Xie, Dragomir Anguelov, Mingxing Tan
Abstract: Current autonomous driving systems rely on specialized models for perceiving and predicting motion, which demonstrate reliable performance in standard conditions. However, generalizing cost-effectively to diverse real-world scenarios remains a significant challenge. To address this, we propose Plug-and-Forecast (PnF), a plug-and-play approach that augments existing motion forecasting models with multimodal large language models (MLLMs). PnF builds on the insight that natural language provides a more effective way to describe and handle complex scenarios, enabling quick adaptation to targeted behaviors. We design prompts to extract structured scene understanding from MLLMs and distill this information into learnable embeddings to augment existing behavior prediction models. Our method leverages the zero-shot reasoning capabilities of MLLMs to achieve significant improvements in motion prediction performance, while requiring no fine-tuning -- making it practical to adopt. We validate our approach on two state-of-the-art motion forecasting models using the Waymo Open Motion Dataset and the nuScenes Dataset, demonstrating consistent performance improvements across both benchmarks.
Authors: Mehdi Zekriyapanah Gashti, Mostafa Mohammadpour, Ghasem Farjamnia
Abstract: Accurate segmentation and classification of white blood cells (WBCs) in microscopic images are essential for diagnosis and monitoring of many hematological disorders, yet remain challenging due to staining variability, complex backgrounds, and class imbalance. In this paper, we introduce a novel Saliency-Guided Cross-Layer Deep Feature Fusion framework (SG-CLDFF) that tightly integrates saliency-driven preprocessing with multi-scale deep feature aggregation to improve both robustness and interpretability for WBC analysis. SG-CLDFF first computes saliency priors to highlight candidate WBC regions and guide subsequent feature extraction. A lightweight hybrid backbone (EfficientSwin-style) produces multi-resolution representations, which are fused by a ResNeXt-CC-inspired cross-layer fusion module to preserve complementary information from shallow and deep layers. The network is trained in a multi-task setup with concurrent segmentation and cell-type classification heads, using class-aware weighted losses and saliency-alignment regularization to mitigate imbalance and suppress background activation. Interpretability is enforced through Grad-CAM visualizations and saliency consistency checks, allowing model decisions to be inspected at the regional level. We validate the framework on standard public benchmarks (BCCD, LISC, ALL-IDB), reporting consistent gains in IoU, F1, and classification accuracy compared to strong CNN and transformer baselines. An ablation study also demonstrates the individual contributions of saliency preprocessing and cross-layer fusion. SG-CLDFF offers a practical and explainable path toward more reliable automated WBC analysis in clinical workflows.
Authors: Amir Gharghabi, Mahdi Hakiminezhad, Maryam Shafaei, Shaghayegh Gharghabi
Abstract: Effortless and ergonomically designed surgical lighting is critical for precision and safety during procedures. However, traditional systems often rely on manual adjustments, leading to surgeon fatigue, neck strain, and inconsistent illumination due to drift and shadowing. To address these challenges, we propose a novel surgical lighting system that leverages the YOLOv11 object detection algorithm to identify a blue marker placed above the target surgical site. A high-power LED light source is then directed to the identified location using two servomotors equipped with tilt-pan brackets. The YOLO model achieves 96.7% mAP@50 on the validation set consisting of annotated images simulating surgical scenes with the blue spherical marker. By automating the lighting process, this machine vision-based solution reduces physical strain on surgeons, improves consistency in illumination, and supports improved surgical outcomes.
Authors: Siran Dai, Qianqian Xu, Peisong Wen, Yang Liu, Qingming Huang
Abstract: In this work, we observe a counterintuitive phenomenon in self-supervised learning (SSL): longer training may impair the performance of dense prediction tasks (e.g., semantic segmentation). We refer to this phenomenon as Self-supervised Dense Degradation (SDD) and demonstrate its consistent presence across sixteen state-of-the-art SSL methods with various losses, architectures, and datasets. When the model performs suboptimally on dense tasks at the end of training, measuring the performance during training becomes essential. However, evaluating dense performance effectively without annotations remains an open challenge. To tackle this issue, we introduce a Dense representation Structure Estimator (DSE), composed of a class-relevance measure and an effective dimensionality measure. The proposed DSE is both theoretically grounded and empirically validated to be closely correlated with the downstream performance. Based on this metric, we introduce a straightforward yet effective model selection strategy and a DSE-based regularization method. Experiments on sixteen SSL methods across four benchmarks confirm that model selection improves mIoU by $3.0\%$ on average with negligible computational cost. Additionally, DSE regularization consistently mitigates the effects of dense degradation. Code is available at https://github.com/EldercatSAM/SSL-Degradation.
Authors: ZhaoYang Han, Qihan Lin, Hao Liang, Bowen Chen, Zhou Liu, Wentao Zhang
Abstract: We introduce \textbf{LongInsightBench}, the first benchmark designed to assess models' ability to understand long videos, with a focus on human language, viewpoints, actions, and other contextual elements, while integrating \textbf{visual, audio, and text} modalities. Our benchmark excels in three key areas: \textbf{a) Long-Duration, Information-Dense Videos:} We carefully select approximately 1,000 videos from open-source datasets FineVideo based on duration limit and the information density of both visual and audio modalities, focusing on content like lectures, interviews, and vlogs, which contain rich language elements. \textbf{b) Diverse and Challenging Task Scenarios:} We have designed six challenging task scenarios, including both Intra-Event and Inter-Event Tasks. \textbf{c) Rigorous and Comprehensive Quality Assurance Pipelines:} We have developed a three-step, semi-automated data quality assurance pipeline to ensure the difficulty and validity of the synthesized questions and answer options. Based on LongInsightBench, we designed a series of experiments. Experimental results shows that Omni-modal models(OLMs) still face challenge in tasks requiring precise temporal localization (T-Loc) and long-range causal inference (CE-Caus). Extended experiments reveal the information loss and processing bias in multi-modal fusion of OLMs. Our dataset and code is available at https://anonymous.4open.science/r/LongInsightBench-910F/.
URLs: https://anonymous.4open.science/r/LongInsightBench-910F/.
Authors: Sangyoon Bae, Jiook Cha
Abstract: We introduce CausalMamba, a scalable framework that addresses fundamental limitations in fMRI-based causal inference: the ill-posed nature of inferring neural causality from hemodynamically distorted BOLD signals and the computational intractability of existing methods like Dynamic Causal Modeling (DCM). Our approach decomposes this complex inverse problem into two tractable stages: BOLD deconvolution to recover latent neural activity, followed by causal graph inference using a novel Conditional Mamba architecture. On simulated data, CausalMamba achieves 37% higher accuracy than DCM. Critically, when applied to real task fMRI data, our method recovers well-established neural pathways with 88% fidelity, whereas conventional approaches fail to identify these canonical circuits in over 99% of subjects. Furthermore, our network analysis of working memory data reveals that the brain strategically shifts its primary causal hub-recruiting executive or salience networks depending on the stimulus-a sophisticated reconfiguration that remains undetected by traditional methods. This work provides neuroscientists with a practical tool for large-scale causal inference that captures both fundamental circuit motifs and flexible network dynamics underlying cognitive function.
Authors: Wei Zhang, Zhanhao Hu, Xiao Li, Xiaopei Zhu, Xiaolin Hu
Abstract: In recent years, adversarial attacks against deep learning-based object detectors in the physical world have attracted much attention. To defend against these attacks, researchers have proposed various defense methods against adversarial patches, a typical form of physically-realizable attack. However, our experiments showed that simply enlarging the patch size could make these defense methods fail. Motivated by this, we evaluated various defense methods against adversarial clothes which have large coverage over the human body. Adversarial clothes provide a good test case for adversarial defense against patch-based attacks because they not only have large sizes but also look more natural than a large patch on humans. Experiments show that all the defense methods had poor performance against adversarial clothes in both the digital world and the physical world. In addition, we crafted a single set of clothes that broke multiple defense methods on Faster R-CNN. The set achieved an Attack Success Rate (ASR) of 96.06% against the undefended detector and over 64.84% ASRs against nine defended models in the physical world, unveiling the common vulnerability of existing adversarial defense methods against adversarial clothes. Code is available at: https://github.com/weiz0823/adv-clothes-break-multiple-defenses.
URLs: https://github.com/weiz0823/adv-clothes-break-multiple-defenses.
Authors: Gyuhwan Park, Kihyun Na, Injung Kim
Abstract: The significance of license plate image restoration goes beyond the preprocessing stage of License Plate Recognition (LPR) systems, as it also serves various purposes, including increasing evidential value, enhancing the clarity of visual interface, and facilitating further utilization of license plate images. We propose a novel diffusion-based framework with character-level guidance, CharDiff, which effectively restores and recognizes severely degraded license plate images captured under realistic conditions. CharDiff leverages fine-grained character-level priors extracted through external segmentation and Optical Character Recognition (OCR) modules tailored for low-quality license plate images. For precise and focused guidance, CharDiff incorporates a novel Character-guided Attention through Region-wise Masking (CHARM) module, which ensures that each character's guidance is restricted to its own region, thereby avoiding interference with other regions. In experiments, CharDiff significantly outperformed the baseline restoration models in both restoration quality and recognition accuracy, achieving a 28% relative reduction in CER on the Roboflow-LP dataset, compared to the best-performing baseline model. These results indicate that the structured character-guided conditioning effectively enhances the robustness of diffusion-based license plate restoration and recognition in practical deployment scenarios.
Authors: Zhaoran Zhao, Xinli Yue, Jianhui Sun, Yuhao Xie, Tao Shao, Liangchao Yao, Fan Xia, Yuetang Deng
Abstract: Image Quality Assessment (IQA) has progressed from scalar quality prediction to more interpretable, human-aligned evaluation paradigms. In this work, we address the emerging challenge of detailed and explainable IQA by proposing iDETEX-a unified multimodal large language model (MLLM) capable of simultaneously performing three key tasks: quality grounding, perception, and description. To facilitate efficient and generalizable training across these heterogeneous subtasks, we design a suite of task-specific offline augmentation modules and a data mixing strategy. These are further complemented by online enhancement strategies to fully exploit multi-sourced supervision. We validate our approach on the large-scale ViDA-UGC benchmark, where iDETEX achieves state-of-the-art performance across all subtasks. Our model ranks first in the ICCV MIPI 2025 Detailed Image Quality Assessment Challenge, demonstrating its effectiveness and robustness in delivering accurate and interpretable quality assessments.
Authors: Jiahao Huo, Mufhumudzi Muthivhi, Terence L. van Zyl, Fredrik Gustafsson
Abstract: Current state-of-the-art Wildlife classification models are trained under the closed world setting. When exposed to unknown classes, they remain overconfident in their predictions. Open-set Recognition (OSR) aims to classify known classes while rejecting unknown samples. Several OSR methods have been proposed to model the closed-set distribution by observing the feature, logit, or softmax probability space. A significant drawback of many existing approaches is the requirement to retrain the pre-trained classification model with the OSR-specific strategy. This study contributes a post-processing OSR method that measures the agreement between the models' features and predicted logits. We propose a probability distribution based on an input's distance to its Nearest Class Mean (NCM). The NCM-based distribution is then compared with the softmax probabilities from the logit space to measure agreement between the NCM and the classification head. Our proposed strategy ranks within the top three on two evaluated datasets, showing consistent performance across the two datasets. In contrast, current state-of-the-art methods excel on a single dataset. We achieve an AUROC of 93.41 and 95.35 for African and Swedish animals. The code can be found https://github.com/Applied-Representation-Learning-Lab/OSR.
URLs: https://github.com/Applied-Representation-Learning-Lab/OSR.
Authors: Jingqian Wu, Shengpeng Xu, Yunbo Jia, Edmund Y. Lam
Abstract: Event cameras offer distinct advantages such as low latency, high dynamic range, and efficient motion capture. However, event-to-video reconstruction (E2V), a fundamental event-based vision task, remains challenging, particularly for reconstructing and recovering semantic information. This is primarily due to the nature of the event camera, as it only captures intensity changes, ignoring static objects and backgrounds, resulting in a lack of semantic information in captured event modality. Further, semantic information plays a crucial role in video and frame reconstruction, yet is often overlooked by existing E2V approaches. To bridge this gap, we propose Semantic-E2VID, an E2V framework that explores the missing visual semantic knowledge in event modality and leverages it to enhance event-to-video reconstruction. Specifically, Semantic-E2VID introduces a cross-modal feature alignment (CFA) module to transfer the robust visual semantics from a frame-based vision foundation model, the Segment Anything Model (SAM), to the event encoder, while aligning the high-level features from distinct modalities. To better utilize the learned semantic feature, we further propose a semantic-aware feature fusion (SFF) block to integrate learned semantics in frame modality to form event representations with rich semantics that can be decoded by the event decoder. Further, to facilitate the reconstruction of semantic information, we propose a novel Semantic Perceptual E2V Supervision that helps the model to reconstruct semantic details by leveraging SAM-generated categorical labels. Extensive experiments demonstrate that Semantic-E2VID significantly enhances frame quality, outperforming state-of-the-art E2V methods across multiple benchmarks. The sample code is included in the supplementary material.
Authors: U. V. B. L Udugama, George Vosselman, Francesco Nex
Abstract: Deploying real-time spatial perception on edge devices requires efficient multi-task models that leverage complementary task information while minimizing computational overhead. This paper introduces Multi-Mono-Hydra (M2H), a novel multi-task learning framework designed for semantic segmentation and depth, edge, and surface normal estimation from a single monocular image. Unlike conventional approaches that rely on independent single-task models or shared encoder-decoder architectures, M2H introduces a Window-Based Cross-Task Attention Module that enables structured feature exchange while preserving task-specific details, improving prediction consistency across tasks. Built on a lightweight ViT-based DINOv2 backbone, M2H is optimized for real-time deployment and serves as the foundation for monocular spatial perception systems supporting 3D scene graph construction in dynamic environments. Comprehensive evaluations show that M2H outperforms state-of-the-art multi-task models on NYUDv2, surpasses single-task depth and semantic baselines on Hypersim, and achieves superior performance on the Cityscapes dataset, all while maintaining computational efficiency on laptop hardware. Beyond benchmarks, M2H is validated on real-world data, demonstrating its practicality in spatial perception tasks.
Authors: Vaggelis Dorovatas, Soroush Seifi, Gunshi Gupta, Rahaf Aljundi
Abstract: Video Large Language Models (Video-LLMs) excel at understanding videos in-context, provided they have full access to the video when answering queries. However, these models face challenges in streaming scenarios where hour-long videos must be processed online, and questions need timely responses. In this work, we propose a training-free approach compatible with standard Video-LLMs, leveraging three key concepts: 1) LLM-informed selection of visual tokens to identify those that the LLM has attended to and contributed to its understanding of each short clip. Our attention-based selection allows us to discard up to ~95% of unimportant visual tokens with minimal performance loss; 2) Recurrent processing of past selected tokens to generate temporally coherent understanding of each processed clip; 3) Caption-based question answering for lightweight and accurate responses. Our method achieves state-of-the-art performance on streaming video benchmarks, striking a balance between efficiency and effectiveness.
Authors: Pawe{\l} Borsukiewicz, Fadi Boutros, Iyiola E. Olatunji, Charles Beumier, Wendk\^uuni C. Ouedraogo, Jacques Klein, Tegawend\'e F. Bissyand\'e
Abstract: The deployment of facial recognition systems has created an ethical dilemma: achieving high accuracy requires massive datasets of real faces collected without consent, leading to dataset retractions and potential legal liabilities under regulations like GDPR. While synthetic facial data presents a promising privacy-preserving alternative, the field lacks comprehensive empirical evidence of its viability. This study addresses this critical gap through extensive evaluation of synthetic facial recognition datasets. We present a systematic literature review identifying 25 synthetic facial recognition datasets (2018-2025), combined with rigorous experimental validation. Our methodology examines seven key requirements for privacy-preserving synthetic data: identity leakage prevention, intra-class variability, identity separability, dataset scale, ethical data sourcing, bias mitigation, and benchmark reliability. Through experiments involving over 10 million synthetic samples, extended by a comparison of results reported on five standard benchmarks, we provide the first comprehensive empirical assessment of synthetic data's capability to replace real datasets. Best-performing synthetic datasets (VariFace, VIGFace) achieve recognition accuracies of 95.67% and 94.91% respectively, surpassing established real datasets including CASIA-WebFace (94.70%). While those images remain private, publicly available alternatives Vec2Face (93.52%) and CemiFace (93.22%) come close behind. Our findings reveal that they ensure proper intra-class variability while maintaining identity separability. Demographic bias analysis shows that, even though synthetic data inherits limited biases, it offers unprecedented control for bias mitigation through generation parameters. These results establish synthetic facial data as a scientifically viable and ethically imperative alternative for facial recognition research.
Authors: Yintao Zhou, Wei Huang, Zhengyu Li, Jing Huang, Meng Pang
Abstract: Parkinson's disease (PD) severity diagnosis is crucial for early detecting potential patients and adopting tailored interventions. Diagnosing PD based on facial expression is grounded in PD patients' "masked face" symptom and gains growing interest recently for its convenience and affordability. However, current facial expression-based approaches often rely on single type of expression which can lead to misdiagnosis, and ignore the class imbalance across different PD stages which degrades the prediction performance. Moreover, most existing methods focus on binary classification (i.e., PD / non-PD) rather than diagnosing the severity of PD. To address these issues, we propose a new facial expression-based method for PD severity diagnosis which integrates multiple facial expression features through attention-based feature fusion. Moreover, we mitigate the class imbalance problem via an adaptive class balancing strategy which dynamically adjusts the contribution of training samples based on their class distribution and classification difficulty. Experimental results demonstrate the promising performance of the proposed method for PD severity diagnosis, as well as the efficacy of attention-based feature fusion and adaptive class balancing.
Authors: Jiajin Tang, Zhengxuan Wei, Ge Zheng, Sibei Yang
Abstract: Humans can perform previously unexperienced interactions with novel objects simply by observing others engage with them. Weakly-supervised affordance grounding mimics this process by learning to locate object regions that enable actions on egocentric images, using exocentric interaction images with image-level annotations. However, extracting affordance knowledge solely from exocentric images and transferring it one-way to egocentric images limits the applicability of previous works in complex interaction scenarios. Instead, this study introduces LoopTrans, a novel closed-loop framework that not only transfers knowledge from exocentric to egocentric but also transfers back to enhance exocentric knowledge extraction. Within LoopTrans, several innovative mechanisms are introduced, including unified cross-modal localization and denoising knowledge distillation, to bridge domain gaps between object-centered egocentric and interaction-centered exocentric images while enhancing knowledge transfer. Experiments show that LoopTrans achieves consistent improvements across all metrics on image and video benchmarks, even handling challenging scenarios where object interaction regions are fully occluded by the human body.
Authors: Dmitrii Galimzianov, Viacheslav Vyshegorodtsev, Ivan Nezhivykh
Abstract: Monitoring the behavior of stalled horses is essential for early detection of health and welfare issues but remains labor-intensive and time-consuming. In this study, we present a prototype vision-based monitoring system that automates the detection and tracking of horses and people inside stables using object detection and multi-object tracking techniques. The system leverages YOLOv11 and BoT-SORT for detection and tracking, while event states are inferred based on object trajectories and spatial relations within the stall. To support development, we constructed a custom dataset annotated with assistance from foundation models CLIP and GroundingDINO. The system distinguishes between five event types and accounts for the camera's blind spots. Qualitative evaluation demonstrated reliable performance for horse-related events, while highlighting limitations in detecting people due to data scarcity. This work provides a foundation for real-time behavioral monitoring in equine facilities, with implications for animal welfare and stable management.
Authors: Shaharyar Ahmed Khan Tareen, Filza Khan Tareen
Abstract: Keypoint detection is the foundation of many computer vision tasks, including image registration, structure-from motion, 3D reconstruction, visual odometry, and SLAM. Traditional detectors (SIFT, SURF, ORB, BRISK, etc.) and learning based methods (SuperPoint, R2D2, LF-Net, D2-Net, etc.) have shown strong performance yet suffer from key limitations: sensitivity to photometric changes, low keypoint density and repeatability, limited adaptability to challenging scenes, and lack of semantic understanding, often failing to prioritize visually important regions. We present DeepDetect, an intelligent, all-in-one, dense keypoint detector that unifies the strengths of classical detectors using deep learning. Firstly, we create ground-truth masks by fusing outputs of 7 keypoint and 2 edge detectors, extracting diverse visual cues from corners and blobs to prominent edges and textures in the images. Afterwards, a lightweight and efficient model: ESPNet, is trained using these masks as labels, enabling DeepDetect to focus semantically on images while producing highly dense keypoints, that are adaptable to diverse and visually degraded conditions. Evaluations on the Oxford Affine Covariant Regions dataset demonstrate that DeepDetect surpasses other detectors in keypoint density, repeatability, and the number of correct matches, achieving maximum values of 0.5143 (average keypoint density), 0.9582 (average repeatability), and 59,003 (correct matches).
Authors: Julien Zouein, Hossein Javidnia, Fran\c{c}ois Piti\'e, Anil Kokaram
Abstract: We repurpose AV1 motion vectors to produce dense sub-pixel correspondences and short tracks filtered by cosine consistency. On short videos, this compressed-domain front end runs comparably to sequential SIFT while using far less CPU, and yields denser matches with competitive pairwise geometry. As a small SfM demo on a 117-frame clip, MV matches register all images and reconstruct 0.46-0.62M points at 0.51-0.53,px reprojection error; BA time grows with match density. These results show compressed-domain correspondences are a practical, resource-efficient front end with clear paths to scaling in full pipelines.
Authors: Qiyuan Guan, Xiang Chen, Guiyue Jin, Jiyu Jin, Shumin Fan, Tianyu Song, Jinshan Pan
Abstract: Compared to daytime image deraining, nighttime image deraining poses significant challenges due to inherent complexities of nighttime scenarios and the lack of high-quality datasets that accurately represent the coupling effect between rain and illumination. In this paper, we rethink the task of nighttime image deraining and contribute a new high-quality benchmark, HQ-NightRain, which offers higher harmony and realism compared to existing datasets. In addition, we develop an effective Color Space Transformation Network (CST-Net) for better removing complex rain from nighttime scenes. Specifically, we propose a learnable color space converter (CSC) to better facilitate rain removal in the Y channel, as nighttime rain is more pronounced in the Y channel compared to the RGB color space. To capture illumination information for guiding nighttime deraining, implicit illumination guidance is introduced enabling the learned features to improve the model's robustness in complex scenarios. Extensive experiments show the value of our dataset and the effectiveness of our method. The source code and datasets are available at https://github.com/guanqiyuan/CST-Net.
Authors: Feng Zhou, Wenkai Guo, Pu Cao, Zhicheng Zhang, Jianqin Yin
Abstract: Sparse-view 3D Gaussian Splatting (3DGS) often overfits to the training views, leading to artifacts like blurring in novel view rendering. Prior work addresses it either by enhancing the initialization (\emph{i.e.}, the point cloud from Structure-from-Motion (SfM)) or by adding training-time constraints (regularization) to the 3DGS optimization. Yet our controlled ablations reveal that initialization is the decisive factor: it determines the attainable performance band in sparse-view 3DGS, while training-time constraints yield only modest within-band improvements at extra cost. Given initialization's primacy, we focus our design there. Although SfM performs poorly under sparse views due to its reliance on feature matching, it still provides reliable seed points. Thus, building on SfM, our effort aims to supplement the regions it fails to cover as comprehensively as possible. Specifically, we design: (i) frequency-aware SfM that improves low-texture coverage via low-frequency view augmentation and relaxed multi-view correspondences; (ii) 3DGS self-initialization that lifts photometric supervision into additional points, compensating SfM-sparse regions with learned Gaussian centers; and (iii) point-cloud regularization that enforces multi-view consistency and uniform spatial coverage through simple geometric/visibility priors, yielding a clean and reliable point cloud. Our experiments on LLFF and Mip-NeRF360 demonstrate consistent gains in sparse-view settings, establishing our approach as a stronger initialization strategy. Code is available at https://github.com/zss171999645/ItG-GS.
Authors: Chenxu Dang, Haiyan Liu, Guangjun Bao, Pei An, Xinyue Tang, Jie Ma, Bingchuan Sun, Yan Wang
Abstract: Semantic occupancy has emerged as a powerful representation in world models for its ability to capture rich spatial semantics. However, most existing occupancy world models rely on static and fixed embeddings or grids, which inherently limit the flexibility of perception. Moreover, their ``in-place classification" over grids exhibits a potential misalignment with the dynamic and continuous nature of real scenarios.In this paper, we propose SparseWorld, a novel 4D occupancy world model that is flexible, adaptive, and efficient, powered by sparse and dynamic queries. We propose a Range-Adaptive Perception module, in which learnable queries are modulated by the ego vehicle states and enriched with temporal-spatial associations to enable extended-range perception. To effectively capture the dynamics of the scene, we design a State-Conditioned Forecasting module, which replaces classification-based forecasting with regression-guided formulation, precisely aligning the dynamic queries with the continuity of the 4D environment. In addition, We specifically devise a Temporal-Aware Self-Scheduling training strategy to enable smooth and efficient training. Extensive experiments demonstrate that SparseWorld achieves state-of-the-art performance across perception, forecasting, and planning tasks. Comprehensive visualizations and ablation studies further validate the advantages of SparseWorld in terms of flexibility, adaptability, and efficiency. The code is available at https://github.com/MSunDYY/SparseWorld.
Authors: Muhammad Umer Ramzan, Ali Zia, Abdelwahed Khamis, Noman Ali, Usman Ali, Wei Xiang
Abstract: Salient object detection (SOD) aims to segment visually prominent regions in images and serves as a foundational task for various computer vision applications. We posit that SOD can now reach near-supervised accuracy without a single pixel-level label, but only when reliable pseudo-masks are available. We revisit the prototype-based line of work and make two key observations. First, boundary pixels and interior pixels obey markedly different geometry; second, the global consistency enforced by optimal transport (OT) is underutilized if prototype quality is weak. To address this, we introduce POTNet, an adaptation of Prototypical Optimal Transport that replaces POT's single k-means step with an entropy-guided dual-clustering head: high-entropy pixels are organized by spectral clustering, low-entropy pixels by k-means, and the two prototype sets are subsequently aligned by OT. This split-fuse-transport design yields sharper, part-aware pseudo-masks in a single forward pass, without handcrafted priors. Those masks supervise a standard MaskFormer-style encoder-decoder, giving rise to AutoSOD, an end-to-end unsupervised SOD pipeline that eliminates SelfMask's offline voting yet improves both accuracy and training efficiency. Extensive experiments on five benchmarks show that AutoSOD outperforms unsupervised methods by up to 26% and weakly supervised methods by up to 36% in F-measure, further narrowing the gap to fully supervised models.
Authors: Yuanli Wu, Long Zhang, Yue Du, Bin Li
Abstract: With the rapid proliferation of video content across social media, surveillance, and education platforms, efficiently summarizing long videos into concise yet semantically faithful surrogates has become increasingly vital. Existing supervised methods achieve strong in-domain accuracy by learning from dense annotations but suffer from high labeling costs and limited cross-dataset generalization, while unsupervised approaches, though label-free, often fail to capture high-level human semantics and fine-grained narrative cues. More recently, zero-shot prompting pipelines have leveraged large language models (LLMs) for training-free video summarization, yet remain highly sensitive to handcrafted prompt templates and dataset-specific score normalization. To overcome these limitations, we introduce a rubric-guided, pseudo-labeled prompting framework that transforms a small subset of ground-truth annotations into high-confidence pseudo labels, which are aggregated into structured, dataset-adaptive scoring rubrics guiding interpretable scene evaluation. During inference, first and last segments are scored based solely on their descriptions, whereas intermediate ones incorporate brief contextual summaries of adjacent scenes to assess narrative progression and redundancy. This contextual prompting enables the LLM to balance local salience and global coherence without parameter tuning. On SumMe and TVSum, our method achieves F1 scores of \textbf{57.58} and \textbf{63.05}, surpassing unsupervised and prior zero-shot baselines while approaching supervised performance. The results demonstrate that rubric-guided pseudo labeling effectively stabilizes LLM-based scoring and establishes a general, interpretable zero-shot paradigm for video summarization.
Authors: Yongshun Zhang, Zhongyi Fan, Yonghang Zhang, Zhangzikang Li, Weifeng Chen, Zhongwei Feng, Chaoyue Wang, Peng Hou, Anxiang Zeng
Abstract: In recent years, large-scale generative models for visual content (\textit{e.g.,} images, videos, and 3D objects/scenes) have made remarkable progress. However, training large-scale video generation models remains particularly challenging and resource-intensive due to cross-modal text-video alignment, the long sequences involved, and the complex spatiotemporal dependencies. To address these challenges, we present a training framework that optimizes four pillars: (i) data processing, (ii) model architecture, (iii) training strategy, and (iv) infrastructure for large-scale video generation models. These optimizations delivered significant efficiency gains and performance improvements across all stages of data preprocessing, video compression, parameter scaling, curriculum-based pretraining, and alignment-focused post-training. Our resulting model, MUG-V 10B, matches recent state-of-the-art video generators overall and, on e-commerce-oriented video generation tasks, surpasses leading open-source baselines in human evaluations. More importantly, we open-source the complete stack, including model weights, Megatron-Core-based large-scale training code, and inference pipelines for video generation and enhancement. To our knowledge, this is the first public release of large-scale video generation training code that exploits Megatron-Core to achieve high training efficiency and near-linear multi-node scaling, details are available in \href{https://github.com/Shopee-MUG/MUG-V}{our webpage}.
Authors: Yovin Yahathugoda, Davide Prezzi, Piyalitt Ittichaiwong, Vicky Goh, Sebastien Ourselin, Michela Antonelli
Abstract: Active Surveillance (AS) is a treatment option for managing low and intermediate-risk prostate cancer (PCa), aiming to avoid overtreatment while monitoring disease progression through serial MRI and clinical follow-up. Accurate prostate segmentation is an important preliminary step for automating this process, enabling automated detection and diagnosis of PCa. However, existing deep-learning segmentation models are often trained on single-time-point and expertly annotated datasets, making them unsuitable for longitudinal AS analysis, where multiple time points and a scarcity of expert labels hinder their effective fine-tuning. To address these challenges, we propose MambaX-Net, a novel semi-supervised, dual-scan 3D segmentation architecture that computes the segmentation for time point t by leveraging the MRI and the corresponding segmentation mask from the previous time point. We introduce two new components: (i) a Mamba-enhanced Cross-Attention Module, which integrates the Mamba block into cross attention to efficiently capture temporal evolution and long-range spatial dependencies, and (ii) a Shape Extractor Module that encodes the previous segmentation mask into a latent anatomical representation for refined zone delination. Moreover, we introduce a semi-supervised self-training strategy that leverages pseudo-labels generated from a pre-trained nnU-Net, enabling effective learning without expert annotations. MambaX-Net was evaluated on a longitudinal AS dataset, and results showed that it significantly outperforms state-of-the-art U-Net and Transformer-based models, achieving superior prostate zone segmentation even when trained on limited and noisy data.
Authors: Nachuan Ma, Zhengfei Song, Qiang Hu, Xiaoyu Tang, Chengxi Zhang, Rui Fan, Lihua Xie
Abstract: Road crack detection is essential for intelligent infrastructure maintenance in smart cities. To reduce reliance on costly pixel-level annotations, we propose WP-CrackNet, an end-to-end weakly-supervised method that trains with only image-level labels for pixel-wise crack detection. WP-CrackNet integrates three components: a classifier generating class activation maps (CAMs), a reconstructor measuring feature inferability, and a detector producing pixel-wise road crack detection results. During training, the classifier and reconstructor alternate in adversarial learning to encourage crack CAMs to cover complete crack regions, while the detector learns from pseudo labels derived from post-processed crack CAMs. This mutual feedback among the three components improves learning stability and detection accuracy. To further boost detection performance, we design a path-aware attention module (PAAM) that fuses high-level semantics from the classifier with low-level structural cues from the reconstructor by modeling spatial and channel-wise dependencies. Additionally, a center-enhanced CAM consistency module (CECCM) is proposed to refine crack CAMs using center Gaussian weighting and consistency constraints, enabling better pseudo-label generation. We create three image-level datasets and extensive experiments show that WP-CrackNet achieves comparable results to supervised methods and outperforms existing weakly-supervised methods, significantly advancing scalable road inspection. The source code package and datasets are available at https://mias.group/WP-CrackNet/.
Authors: Kaichen Zhou, Yuhan Wang, Grace Chen, Xinhai Chang, Gaspard Beaudouin, Fangneng Zhan, Paul Pu Liang, Mengyu Wang
Abstract: Recent 3D feed-forward models, such as the Visual Geometry Grounded Transformer (VGGT), have shown strong capability in inferring 3D attributes of static scenes. However, since they are typically trained on static datasets, these models often struggle in real-world scenarios involving complex dynamic elements, such as moving humans or deformable objects like umbrellas. To address this limitation, we introduce PAGE-4D, a feedforward model that extends VGGT to dynamic scenes, enabling camera pose estimation, depth prediction, and point cloud reconstruction -- all without post-processing. A central challenge in multi-task 4D reconstruction is the inherent conflict between tasks: accurate camera pose estimation requires suppressing dynamic regions, while geometry reconstruction requires modeling them. To resolve this tension, we propose a dynamics-aware aggregator that disentangles static and dynamic information by predicting a dynamics-aware mask -- suppressing motion cues for pose estimation while amplifying them for geometry reconstruction. Extensive experiments show that PAGE-4D consistently outperforms the original VGGT in dynamic scenarios, achieving superior results in camera pose estimation, monocular and video depth estimation, and dense point map reconstruction.
Authors: Chuhong Wang, Hua Li, Chongyi Li, Huazhong Liu, Xiongxin Tang, Sam Kwong
Abstract: With the development of underwater exploration and marine protection, underwater vision tasks are widespread. Due to the degraded underwater environment, characterized by color distortion, low contrast, and blurring, camouflaged instance segmentation (CIS) faces greater challenges in accurately segmenting objects that blend closely with their surroundings. Traditional camouflaged instance segmentation methods, trained on terrestrial-dominated datasets with limited underwater samples, may exhibit inadequate performance in underwater scenes. To address these issues, we introduce the first underwater camouflaged instance segmentation (UCIS) dataset, abbreviated as UCIS4K, which comprises 3,953 images of camouflaged marine organisms with instance-level annotations. In addition, we propose an Underwater Camouflaged Instance Segmentation network based on Segment Anything Model (UCIS-SAM). Our UCIS-SAM includes three key modules. First, the Channel Balance Optimization Module (CBOM) enhances channel characteristics to improve underwater feature learning, effectively addressing the model's limited understanding of underwater environments. Second, the Frequency Domain True Integration Module (FDTIM) is proposed to emphasize intrinsic object features and reduce interference from camouflage patterns, enhancing the segmentation performance of camouflaged objects blending with their surroundings. Finally, the Multi-scale Feature Frequency Aggregation Module (MFFAM) is designed to strengthen the boundaries of low-contrast camouflaged instances across multiple frequency bands, improving the model's ability to achieve more precise segmentation of camouflaged objects. Extensive experiments on the proposed UCIS4K and public benchmarks show that our UCIS-SAM outperforms state-of-the-art approaches.
Authors: Shuyuan Zhang, Chenhan Jiang, Zuoou Li, Jiankang Deng
Abstract: 3D generation from natural language offers significant potential to reduce expert manual modeling efforts and enhance accessibility to 3D assets. However, existing methods often yield unstructured meshes and exhibit poor interactivity, making them impractical for artistic workflows. To address these limitations, we represent 3D assets as shape programs and introduce ShapeCraft, a novel multi-agent framework for text-to-3D generation. At its core, we propose a Graph-based Procedural Shape (GPS) representation that decomposes complex natural language into a structured graph of sub-tasks, thereby facilitating accurate LLM comprehension and interpretation of spatial relationships and semantic shape details. Specifically, LLM agents hierarchically parse user input to initialize GPS, then iteratively refine procedural modeling and painting to produce structured, textured, and interactive 3D assets. Qualitative and quantitative experiments demonstrate ShapeCraft's superior performance in generating geometrically accurate and semantically rich 3D assets compared to existing LLM-based agents. We further show the versatility of ShapeCraft through examples of animated and user-customized editing, highlighting its potential for broader interactive applications.
Authors: Siqi Chen, Shanyue Guan
Abstract: The advancement of UAV technology has enabled efficient, non-contact structural health monitoring. Combined with photogrammetry, UAVs can capture high-resolution scans and reconstruct detailed 3D models of infrastructure. However, a key challenge remains in segmenting specific structural components from these models-a process traditionally reliant on time-consuming and error-prone manual labeling. To address this issue, we propose a machine learning-based framework for automated segmentation of 3D point clouds. Our approach uses the complementary strengths of real-world UAV-scanned point clouds and synthetic data generated from Building Information Modeling (BIM) to overcome the limitations associated with manual labeling. Validation on a railroad track dataset demonstrated high accuracy in identifying and segmenting major components such as rails and crossties. Moreover, by using smaller-scale datasets supplemented with BIM data, the framework significantly reduced training time while maintaining reasonable segmentation accuracy. This automated approach improves the precision and efficiency of 3D infrastructure model segmentation and advances the integration of UAV and BIM technologies in structural health monitoring and infrastructure management.
Authors: Jia Guo, Shuai Lu, Lei Fan, Zelin Li, Donglin Di, Yang Song, Weihang Zhang, Wenbing Zhu, Hong Yan, Fang Chen, Huiqi Li, Hongen Liao
Abstract: Unsupervised anomaly detection (UAD) has evolved from building specialized single-class models to unified multi-class models, yet existing multi-class models significantly underperform the most advanced one-for-one counterparts. Moreover, the field has fragmented into specialized methods tailored to specific scenarios (multi-class, 3D, few-shot, etc.), creating deployment barriers and highlighting the need for a unified solution. In this paper, we present Dinomaly2, the first unified framework for full-spectrum image UAD, which bridges the performance gap in multi-class models while seamlessly extending across diverse data modalities and task settings. Guided by the "less is more" philosophy, we demonstrate that the orchestration of five simple element achieves superior performance in a standard reconstruction-based framework. This methodological minimalism enables natural extension across diverse tasks without modification, establishing that simplicity is the foundation of true universality. Extensive experiments on 12 UAD benchmarks demonstrate Dinomaly2's full-spectrum superiority across multiple modalities (2D, multi-view, RGB-3D, RGB-IR), task settings (single-class, multi-class, inference-unified multi-class, few-shot) and application domains (industrial, biological, outdoor). For example, our multi-class model achieves unprecedented 99.9% and 99.3% image-level (I-) AUROC on MVTec-AD and VisA respectively. For multi-view and multi-modal inspection, Dinomaly2 demonstrates state-of-the-art performance with minimum adaptations. Moreover, using only 8 normal examples per class, our method surpasses previous full-shot models, achieving 98.7% and 97.4% I-AUROC on MVTec-AD and VisA. The combination of minimalistic design, computational scalability, and universal applicability positions Dinomaly2 as a unified solution for the full spectrum of real-world anomaly detection applications.
Authors: Fr\'ed\'eric LIN (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France), Biruk Abere Ambaw (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France), Adrian Popescu (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France), Hejer Ammar (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France), Romaric Audigier (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France), Herv\'e Le Borgne (Universit\'e Paris-Saclay, CEA, List, F-91120, Palaiseau, France)
Abstract: AI systems must adapt to evolving visual environments, especially in domains where object appearances change over time. We introduce Car Models in Time (CaMiT), a fine-grained dataset capturing the temporal evolution of car models, a representative class of technological artifacts. CaMiT includes 787K labeled samples of 190 car models (2007-2023) and 5.1M unlabeled samples (2005-2023), supporting both supervised and self-supervised learning. Static pretraining on in-domain data achieves competitive performance with large-scale generalist models while being more resource-efficient, yet accuracy declines when models are tested across years. To address this, we propose a time-incremental classification setting, a realistic continual learning scenario with emerging, evolving, and disappearing classes. We evaluate two strategies: time-incremental pretraining, which updates the backbone, and time-incremental classifier learning, which updates only the final layer, both improving temporal robustness. Finally, we explore time-aware image generation that leverages temporal metadata during training, yielding more realistic outputs. CaMiT offers a rich benchmark for studying temporal adaptation in fine-grained visual recognition and generation.
Authors: Zexian Huang, Mashnoon Islam, Brian Armstrong, Kourosh Khoshelham, Martin Tomko
Abstract: Dry-stone walls hold significant heritage and environmental value. Mapping these structures is essential for ecosystem preservation and wildfire management in Australia. Yet, many walls remain unidentified due to their inaccessibility and the high cost of manual mapping. Deep learning-based segmentation offers a scalable solution, but two major challenges persist: (1) visual occlusion of low-lying walls by dense vegetation, and (2) limited labeled data for supervised training. We propose DINO-CV, a segmentation framework for automatic mapping of low-lying dry-stone walls using high-resolution Airborne LiDAR-derived digital elevation models (DEMs). DEMs overcome visual occlusion by capturing terrain structures hidden beneath vegetation, enabling analysis of structural rather than spectral cues. DINO-CV introduces a self-supervised cross-view pre-training strategy based on knowledge distillation to mitigate data scarcity. It learns invariant visual and geometric representations across multiple DEM derivatives, supporting various vision backbones including ResNet, Wide ResNet, and Vision Transformers. Applied to the UNESCO World Heritage cultural landscape of Budj Bim, Victoria, the method identifies one of Australia's densest collections of colonial dry-stone walls beyond Indigenous heritage contexts. DINO-CV achieves a mean Intersection over Union (mIoU) of 68.6% on test areas and maintains 63.8% mIoU when fine-tuned with only 10% labeled data. These results demonstrate the potential of self-supervised learning on high-resolution DEM derivatives for automated dry-stone wall mapping in vegetated and heritage-rich environments with scarce annotations.
Authors: S\'ebastien Thuau, Siba Haidar, Ayush Bajracharya, Rachid Chelouah
Abstract: We examine frugal federated learning approaches to violence detection by comparing two complementary strategies: (i) zero-shot and federated fine-tuning of vision-language models (VLMs), and (ii) personalized training of a compact 3D convolutional neural network (CNN3D). Using LLaVA-7B and a 65.8M parameter CNN3D as representative cases, we evaluate accuracy, calibration, and energy usage under realistic non-IID settings. Both approaches exceed 90% accuracy. CNN3D slightly outperforms Low-Rank Adaptation(LoRA)-tuned VLMs in ROC AUC and log loss, while using less energy. VLMs remain favorable for contextual reasoning and multimodal inference. We quantify energy and CO$_2$ emissions across training and inference, and analyze sustainability trade-offs for deployment. To our knowledge, this is the first comparative study of LoRA-tuned vision-language models and personalized CNNs for federated violence detection, with an emphasis on energy efficiency and environmental metrics. These findings support a hybrid model: lightweight CNNs for routine classification, with selective VLM activation for complex or descriptive scenarios. The resulting framework offers a reproducible baseline for responsible, resource-aware AI in video surveillance, with extensions toward real-time, multimodal, and lifecycle-aware systems.
Authors: Ling Liu, Jun Tian, Li Yi
Abstract: 4D panoptic segmentation in a streaming setting is critical for highly dynamic environments, such as evacuating dense crowds and autonomous driving in complex scenarios, where real-time, fine-grained perception within a constrained time budget is essential. In this paper, we introduce 4DSegStreamer, a novel framework that employs a Dual-Thread System to efficiently process streaming frames. The framework is general and can be seamlessly integrated into existing 3D and 4D segmentation methods to enable real-time capability. It also demonstrates superior robustness compared to existing streaming perception approaches, particularly under high FPS conditions. The system consists of a predictive thread and an inference thread. The predictive thread leverages historical motion and geometric information to extract features and forecast future dynamics. The inference thread ensures timely prediction for incoming frames by aligning with the latest memory and compensating for ego-motion and dynamic object movements. We evaluate 4DSegStreamer on the indoor HOI4D dataset and the outdoor SemanticKITTI and nuScenes datasets. Comprehensive experiments demonstrate the effectiveness of our approach, particularly in accurately predicting dynamic objects in complex scenes.
Authors: Yuandong Pu, Le Zhuo, Songhao Han, Jinbo Xing, Kaiwen Zhu, Shuo Cao, Bin Fu, Si Liu, Hongsheng Li, Yu Qiao, Wenlong Zhang, Xi Chen, Yihao Liu
Abstract: Image editing has achieved remarkable progress recently. Modern editing models could already follow complex instructions to manipulate the original content. However, beyond completing the editing instructions, the accompanying physical effects are the key to the generation realism. For example, removing an object should also remove its shadow, reflections, and interactions with nearby objects. Unfortunately, existing models and benchmarks mainly focus on instruction completion but overlook these physical effects. So, at this moment, how far are we from physically realistic image editing? To answer this, we introduce PICABench, which systematically evaluates physical realism across eight sub-dimension (spanning optics, mechanics, and state transitions) for most of the common editing operations (add, remove, attribute change, etc). We further propose the PICAEval, a reliable evaluation protocol that uses VLM-as-a-judge with per-case, region-level human annotations and questions. Beyond benchmarking, we also explore effective solutions by learning physics from videos and construct a training dataset PICA-100K. After evaluating most of the mainstream models, we observe that physical realism remains a challenging problem with large rooms to explore. We hope that our benchmark and proposed solutions can serve as a foundation for future work moving from naive content editing toward physically consistent realism.
Authors: Xinwei Zhang, Hu Chen, Zhe Yuan, Sukun Tian, Peng Feng
Abstract: Foundation models for medical image segmentation have achieved remarkable performance. Adaptive fine-tuning of natural image segmentation foundation models is crucial for medical image segmentation tasks. However, some limitations exist in existing fine-tuning methods: 1) insufficient representation of high-level features and 2) the fine-tuning process disrupts the structural integrity of pretrained weights. Inspired by these critical problems, we propose an intelligent communication mixture-of-experts boosted-medical image segmentation foundation model, named IC-MoE, with twofold ideas: 1) We construct basic experts, semantic experts, and adaptive experts. Moreover, we implement a pixel probability adaptive voting strategy, which enables expert selection and fusion through label consistency and load balancing. This approach preliminarily enhances the representation capability of high-level features while preserving the structural integrity of pretrained weights. 2) We propose a semantic-guided contrastive learning method to address the issue of weak supervision in contrastive learning. This method further enhances the representation capability of high-level features while preserving the structural integrity of pretrained weights. Extensive experiments across three public medical image segmentation datasets demonstrate that the IC-MoE outperforms other SOTA models. Consequently, the proposed IC-MoE effectively supplements foundational medical image segmentation models with high-level features and pretrained structural integrity. We also validate the superior generalizability of the IC-MoE across diverse medical image segmentation scenarios.
Authors: Min Cao, Xinyu Zhou, Ding Jiang, Bo Du, Mang Ye, Min Zhang
Abstract: Text-to-image person retrieval (TIPR) aims to identify the target person using textual descriptions, facing challenge in modality heterogeneity. Prior works have attempted to address it by developing cross-modal global or local alignment strategies. However, global methods typically overlook fine-grained cross-modal differences, whereas local methods require prior information to explore explicit part alignments. Additionally, current methods are English-centric, restricting their application in multilingual contexts. To alleviate these issues, we pioneer a multilingual TIPR task by developing a multilingual TIPR benchmark, for which we leverage large language models for initial translations and refine them by integrating domain-specific knowledge. Correspondingly, we propose Bi-IRRA: a Bidirectional Implicit Relation Reasoning and Aligning framework to learn alignment across languages and modalities. Within Bi-IRRA, a bidirectional implicit relation reasoning module enables bidirectional prediction of masked image and text, implicitly enhancing the modeling of local relations across languages and modalities, a multi-dimensional global alignment module is integrated to bridge the modality heterogeneity. The proposed method achieves new state-of-the-art results on all multilingual TIPR datasets. Data and code are presented in https://github.com/Flame-Chasers/Bi-IRRA.
Authors: Taichi Liu, Zhenyu Wang, Ruofeng Liu, Guang Wang, Desheng Zhang
Abstract: Recent advancements in 3D object detection and novel category detection have made significant progress, yet research on learning generalized 3D objectness remains insufficient. In this paper, we delve into learning open-world 3D objectness, which focuses on detecting all objects in a 3D scene, including novel objects unseen during training. Traditional closed-set 3D detectors struggle to generalize to open-world scenarios, while directly incorporating 3D open-vocabulary models for open-world ability struggles with vocabulary expansion and semantic overlap. To achieve generalized 3D object discovery, We propose OP3Det, a class-agnostic Open-World Prompt-free 3D Detector to detect any objects within 3D scenes without relying on hand-crafted text prompts. We introduce the strong generalization and zero-shot capabilities of 2D foundation models, utilizing both 2D semantic priors and 3D geometric priors for class-agnostic proposals to broaden 3D object discovery. Then, by integrating complementary information from point cloud and RGB image in the cross-modal mixture of experts, OP3Det dynamically routes uni-modal and multi-modal features to learn generalized 3D objectness. Extensive experiments demonstrate the extraordinary performance of OP3Det, which significantly surpasses existing open-world 3D detectors by up to 16.0% in AR and achieves a 13.5% improvement compared to closed-world 3D detectors.
Authors: Aleksandr Oganov, Ilya Bykov, Eva Neudachina, Mishan Aliev, Alexander Tolmachev, Alexander Sidorov, Aleksandr Zuev, Andrey Okhotin, Denis Rakitin, Aibek Alanov
Abstract: While diffusion models achieve state-of-the-art generation quality, they still suffer from computationally expensive sampling. Recent works address this issue with gradient-based optimization methods that distill a few-step ODE diffusion solver from the full sampling process, reducing the number of function evaluations from dozens to just a few. However, these approaches often rely on intricate training techniques and do not explicitly focus on preserving fine-grained details. In this paper, we introduce the Generalized Solver: a simple parameterization of the ODE sampler that does not require additional training tricks and improves quality over existing approaches. We further combine the original distillation loss with adversarial training, which mitigates artifacts and enhances detail fidelity. We call the resulting method the Generalized Adversarial Solver and demonstrate its superior performance compared to existing solver training methods under similar resource constraints. Code is available at https://github.com/3145tttt/GAS.
Authors: Walter Simoncini, Michael Dorkenwald, Tijmen Blankevoort, Cees G. M. Snoek, Yuki M. Asano
Abstract: Vision foundation models achieve remarkable performance but are only available in a limited set of pre-determined sizes, forcing sub-optimal deployment choices under real-world constraints. We introduce SnapViT: Single-shot network approximation for pruned Vision Transformers, a new post-pretraining structured pruning method that enables elastic inference across a continuum of compute budgets. Our approach efficiently combines gradient information with cross-network structure correlations, approximated via an evolutionary algorithm, does not require labeled data, generalizes to models without a classification head, and is retraining-free. Experiments on DINO, SigLIPv2, DeIT, and AugReg models demonstrate superior performance over state-of-the-art methods across various sparsities, requiring less than five minutes on a single A100 GPU to generate elastic models that can be adjusted to any computational budget. Our key contributions include an efficient pruning strategy for pretrained Vision Transformers, a novel evolutionary approximation of Hessian off-diagonal structures, and a self-supervised importance scoring mechanism that maintains strong performance without requiring retraining or labels. Code and pruned models are available at: https://elastic.ashita.nl/
Authors: Mhd Adnan Albani, Riad Sonbol
Abstract: Parkinson's disease (PD) is a neurodegenerative disease affecting about 1% of people over the age of 60, causing motor impairments that impede hand coordination activities such as writing and drawing. Many approaches have tried to support early detection of Parkinson's disease based on hand-drawn images; however, we identified two major limitations in the related works: (1) the lack of sufficient datasets, (2) the robustness when dealing with unseen patient data. In this paper, we propose a new approach to detect Parkinson's disease that consists of two stages: The first stage classifies based on their drawing type(circle, meander, spiral), and the second stage extracts the required features from the images and detects Parkinson's disease. We overcame the previous two limitations by applying a chunking strategy where we divide each image into 2x2 chunks. Each chunk is processed separately when extracting features and recognizing Parkinson's disease indicators. To make the final classification, an ensemble method is used to merge the decisions made from each chunk. Our evaluation shows that our proposed approach outperforms the top performing state-of-the-art approaches, in particular on unseen patients. On the NewHandPD dataset our approach, it achieved 97.08% accuracy for seen patients and 94.91% for unseen patients, our proposed approach maintained a gap of only 2.17 percentage points, compared to the 4.76-point drop observed in prior work.
Authors: Suqiang Ma, Subhadeep Sengupta, Yao Lee, Beikang Gu, Xianyan Chen, Xianqiao Wang, Yang Liu, Mengjia Xu, Galit H. Frydman, He Li
Abstract: Circulating blood cell clusters (CCCs) containing red blood cells (RBCs), white blood cells(WBCs), and platelets are significant biomarkers linked to conditions like thrombosis, infection, and inflammation. Flow cytometry, paired with fluorescence staining, is commonly used to analyze these cell clusters, revealing cell morphology and protein profiles. While computational approaches based on machine learning have advanced the automatic analysis of single-cell flow cytometry images, there is a lack of effort to build tools to automatically analyze images containing CCCs. Unlike single cells, cell clusters often exhibit irregular shapes and sizes. In addition, these cell clusters often consist of heterogeneous cell types, which require multi-channel staining to identify the specific cell types within the clusters. This study introduces a new computational framework for analyzing CCC images and identifying cell types within clusters. Our framework uses a two-step analysis strategy. First, it categorizes images into cell cluster and non-cluster groups by fine-tuning the You Only Look Once(YOLOv11) model, which outperforms traditional convolutional neural networks (CNNs), Vision Transformers (ViT). Then, it identifies cell types by overlaying cluster contours with regions from multi-channel fluorescence stains, enhancing accuracy despite cell debris and staining artifacts. This approach achieved over 95% accuracy in both cluster classification and phenotype identification. In summary, our automated framework effectively analyzes CCC images from flow cytometry, leveraging both bright-field and fluorescence data. Initially tested on blood cells, it holds potential for broader applications, such as analyzing immune and tumor cell clusters, supporting cellular research across various diseases.
Authors: Zhiqiang Teng, Beibei Lin, Tingting Chen, Zifeng Yuan, Xuanyi Li, Xuanyu Zhang, Shunli Zhang
Abstract: 3D Gaussian Splatting (3DGS) under raindrop conditions suffers from severe occlusions and optical distortions caused by raindrop contamination on the camera lens, substantially degrading reconstruction quality. Existing benchmarks typically evaluate 3DGS using synthetic raindrop images with known camera poses (constrained images), assuming ideal conditions. However, in real-world scenarios, raindrops often interfere with accurate camera pose estimation and point cloud initialization. Moreover, a significant domain gap between synthetic and real raindrops further impairs generalization. To tackle these issues, we introduce RaindropGS, a comprehensive benchmark designed to evaluate the full 3DGS pipeline-from unconstrained, raindrop-corrupted images to clear 3DGS reconstructions. Specifically, the whole benchmark pipeline consists of three parts: data preparation, data processing, and raindrop-aware 3DGS evaluation, including types of raindrop interference, camera pose estimation and point cloud initialization, single image rain removal comparison, and 3D Gaussian training comparison. First, we collect a real-world raindrop reconstruction dataset, in which each scene contains three aligned image sets: raindrop-focused, background-focused, and rain-free ground truth, enabling a comprehensive evaluation of reconstruction quality under different focus conditions. Through comprehensive experiments and analyses, we reveal critical insights into the performance limitations of existing 3DGS methods on unconstrained raindrop images and the varying impact of different pipeline components: the impact of camera focus position on 3DGS reconstruction performance, and the interference caused by inaccurate pose and point cloud initialization on reconstruction. These insights establish clear directions for developing more robust 3DGS methods under raindrop conditions.
Authors: Yaning Pan, Zekun Wang, Qianqian Xie, Yongqian Wen, Yuanxing Zhang, Guohui Zhang, Haoxuan Hu, Zhiyu Pan, Yibing Huang, Zhidong Gan, Yonghong Lin, An Ping, Tianhao Peng, Jiaheng Liu
Abstract: The recent development of Multimodal Large Language Models (MLLMs) has significantly advanced AI's ability to understand visual modalities. However, existing evaluation benchmarks remain limited to single-turn question answering, overlooking the complexity of multi-turn dialogues in real-world scenarios. To bridge this gap, we introduce MT-Video-Bench, a holistic video understanding benchmark for evaluating MLLMs in multi-turn dialogues. Specifically, our MT-Video-Bench mainly assesses six core competencies that focus on perceptivity and interactivity, encompassing 987 meticulously curated multi-turn dialogues from diverse domains. These capabilities are rigorously aligned with real-world applications, such as interactive sports analysis and multi-turn video-based intelligent tutoring. With MT-Video-Bench, we extensively evaluate various state-of-the-art open-source and closed-source MLLMs, revealing their significant performance discrepancies and limitations in handling multi-turn video dialogues. The benchmark will be publicly available to foster future research.
Authors: Matheus Ramos Parracho
Abstract: Automated signature verification is a critical biometric technique used in banking, identity authentication, and legal documentation. Despite the notable progress achieved by deep learning methods, most approaches in offline signature verification still struggle to generalize across datasets, as variations in handwriting styles and acquisition protocols often degrade performance. This study investigates feature learning strategies for signature forgery detection, focusing on improving cross-dataset generalization -- that is, model robustness when trained on one dataset and tested on another. Using three public benchmarks -- CEDAR, ICDAR, and GPDS Synthetic -- two experimental pipelines were developed: one based on raw signature images and another employing a preprocessing method referred to as shell preprocessing. Several behavioral patterns were identified and analyzed; however, no definitive superiority between the two approaches was established. The results show that the raw-image model achieved higher performance across benchmarks, while the shell-based model demonstrated promising potential for future refinement toward robust, cross-domain signature verification.
Authors: Aaron Appelle, Jerome P. Lynch
Abstract: Recent high-performing image-to-video (I2V) models based on variants of the diffusion transformer (DiT) have displayed remarkable inherent world-modeling capabilities by virtue of training on large scale video datasets. We investigate whether these models can generate realistic pedestrian movement patterns in crowded public scenes. Our framework conditions I2V models on keyframes extracted from pedestrian trajectory benchmarks, then evaluates their trajectory prediction performance using quantitative measures of pedestrian dynamics.
Authors: Timur Ismagilov, Shakaiba Majeed, Michael Milford, Tan Viet Tuyen Nguyen, Sarvapali D. Ramchurn, Shoaib Ehsan
Abstract: We address multi-reference visual place recognition (VPR), where reference sets captured under varying conditions are used to improve localisation performance. While deep learning with large-scale training improves robustness, increasing data diversity and model complexity incur extensive computational cost during training and deployment. Descriptor-level fusion via voting or aggregation avoids training, but often targets multi-sensor setups or relies on heuristics with limited gains under appearance and viewpoint change. We propose a training-free, descriptor-agnostic approach that jointly models places using multiple reference descriptors via matrix decomposition into basis representations, enabling projection-based residual matching. We also introduce SotonMV, a structured benchmark for multi-viewpoint VPR. On multi-appearance data, our method improves Recall@1 by up to ~18% over single-reference and outperforms multi-reference baselines across appearance and viewpoint changes, with gains of ~5% on unstructured data, demonstrating strong generalisation while remaining lightweight.
Authors: Md. Enamul Atiq, Shaikh Anowarul Fattah
Abstract: Skin cancer is a life-threatening disease where early detection significantly improves patient outcomes. Automated diagnosis from dermoscopic images is challenging due to high intra-class variability and subtle inter-class differences. Many deep learning models operate as "black boxes," limiting clinical trust. In this work, we propose a dual-encoder attention-based framework that leverages both segmented lesions and clinical metadata to enhance skin lesion classification in terms of both accuracy and interpretability. A novel Deep-UNet architecture with Dual Attention Gates (DAG) and Atrous Spatial Pyramid Pooling (ASPP) is first employed to segment lesions. The classification stage uses two DenseNet201 encoders-one on the original image and another on the segmented lesion whose features are fused via multi-head cross-attention. This dual-input design guides the model to focus on salient pathological regions. In addition, a transformer-based module incorporates patient metadata (age, sex, lesion site) into the prediction. We evaluate our approach on the HAM10000 dataset and the ISIC 2018 and 2019 challenges. The proposed method achieves state-of-the-art segmentation performance and significantly improves classification accuracy and average AUC compared to baseline models. To validate our model's reliability, we use Gradient-weighted Class Activation Mapping (Grad-CAM) to generate heatmaps. These visualizations confirm that our model's predictions are based on the lesion area, unlike models that rely on spurious background features. These results demonstrate that integrating precise lesion segmentation and clinical data with attention-based fusion leads to a more accurate and interpretable skin cancer classification model.
Authors: Samir Khaki, Junxian Guo, Jiaming Tang, Shang Yang, Yukang Chen, Konstantinos N. Plataniotis, Yao Lu, Song Han, Zhijian Liu
Abstract: Vision Language Models (VLMs) have rapidly advanced in integrating visual and textual reasoning, powering applications across high-resolution image understanding, long-video analysis, and multi-turn conversation. However, their scalability remains limited by the growing number of visual tokens that dominate inference latency. We present SparseVILA, a new paradigm for efficient VLM inference that decouples visual sparsity across the prefilling and decoding stages. SparseVILA distributes sparsity across stages by pruning redundant visual tokens during prefill and retrieving only query-relevant tokens during decoding. This decoupled design matches leading prefill pruning methods while preserving multi-turn fidelity by retaining most of the visual cache so that query-aware tokens can be retrieved at each conversation round. Built on an AWQ-optimized inference pipeline, SparseVILA achieves up to 4.0 times faster prefilling, 2.5 times faster decoding, and an overall 2.6 times end-to-end speedup on long-context video tasks -- while improving accuracy on document-understanding and reasoning tasks. By decoupling query-agnostic pruning and query-aware retrieval, SparseVILA establishes a new direction for efficient multimodal inference, offering a training-free, architecture-agnostic framework for accelerating large VLMs without sacrificing capability.
Authors: Yuhao Yang, Zhen Yang, Zi-Yi Dou, Anh Nguyen, Keen You, Omar Attia, Andrew Szot, Michael Feng, Ram Ramrakhya, Alexander Toshev, Chao Huang, Yinfei Yang, Zhe Gan
Abstract: Multimodal agents for computer use rely exclusively on primitive actions (click, type, scroll) that require accurate visual grounding and lengthy execution chains, leading to cascading failures and performance bottlenecks. While other agents leverage rich programmatic interfaces (APIs, MCP servers, tools), computer-use agents (CUAs) remain isolated from these capabilities. We present UltraCUA, a foundation model that bridges this gap through hybrid action -- seamlessly integrating GUI primitives with high-level programmatic tool calls. To achieve this, our approach comprises four key components: (1) an automated pipeline that scales programmatic tools from software documentation, open-source repositories, and code generation; (2) a synthetic data engine producing over 17,000 verifiable tasks spanning real-world computer-use scenarios; (3) a large-scale high-quality hybrid action trajectory collection with both low-level GUI actions and high-level programmatic tool calls; and (4) a two-stage training pipeline combining supervised fine-tuning with online reinforcement learning, enabling strategic alternation between low-level and high-level actions. Experiments with our 7B and 32B models demonstrate substantial improvements over state-of-the-art agents. On OSWorld, UltraCUA models achieve an average 22% relative improvement over base models, while being 11% faster in terms of steps. Out-of-domain evaluation on WindowsAgentArena shows our model reaches 21.7% success rate, outperforming baselines trained on Windows data. The hybrid action mechanism proves critical, reducing error propagation while maintaining execution efficiency.
Authors: Jiale Cheng, Yusen Liu, Xinyu Zhang, Yulin Fei, Wenyi Hong, Ruiliang Lyu, Weihan Wang, Zhe Su, Xiaotao Gu, Xiao Liu, Yushi Bai, Jie Tang, Hongning Wang, Minlie Huang
Abstract: Large language models (LLMs) increasingly rely on long-context modeling for tasks such as document understanding, code analysis, and multi-step reasoning. However, scaling context windows to the million-token level brings prohibitive computational and memory costs, limiting the practicality of long-context LLMs. In this work, we take a different perspective-visual context scaling-to tackle this challenge. Instead of extending token-based sequences, we propose Glyph, a framework that renders long texts into images and processes them with vision-language models (VLMs). This approach substantially compresses textual input while preserving semantic information, and we further design an LLM-driven genetic search to identify optimal visual rendering configurations for balancing accuracy and compression. Through extensive experiments, we demonstrate that our method achieves 3-4x token compression while maintaining accuracy comparable to leading LLMs such as Qwen3-8B on various long-context benchmarks. This compression also leads to around 4x faster prefilling and decoding, and approximately 2x faster SFT training. Furthermore, under extreme compression, a 128K-context VLM could scale to handle 1M-token-level text tasks. In addition, the rendered text data benefits real-world multimodal tasks, such as document understanding. Our code and model are released at https://github.com/thu-coai/Glyph.
Authors: Zixin Yin, Ling-Hao Chen, Lionel Ni, Xili Dai
Abstract: Recent advances in training-free attention control methods have enabled flexible and efficient text-guided editing capabilities for existing generation models. However, current approaches struggle to simultaneously deliver strong editing strength while preserving consistency with the source. This limitation becomes particularly critical in multi-round and video editing, where visual errors can accumulate over time. Moreover, most existing methods enforce global consistency, which limits their ability to modify individual attributes such as texture while preserving others, thereby hindering fine-grained editing. Recently, the architectural shift from U-Net to MM-DiT has brought significant improvements in generative performance and introduced a novel mechanism for integrating text and vision modalities. These advancements pave the way for overcoming challenges that previous methods failed to resolve. Through an in-depth analysis of MM-DiT, we identify three key insights into its attention mechanisms. Building on these, we propose ConsistEdit, a novel attention control method specifically tailored for MM-DiT. ConsistEdit incorporates vision-only attention control, mask-guided pre-attention fusion, and differentiated manipulation of the query, key, and value tokens to produce consistent, prompt-aligned edits. Extensive experiments demonstrate that ConsistEdit achieves state-of-the-art performance across a wide range of image and video editing tasks, including both structure-consistent and structure-inconsistent scenarios. Unlike prior methods, it is the first approach to perform editing across all inference steps and attention layers without handcraft, significantly enhancing reliability and consistency, which enables robust multi-round and multi-region editing. Furthermore, it supports progressive adjustment of structural consistency, enabling finer control.
Authors: Lunchen Xie, Zehua He, Qingjiang Shi
Abstract: Personalized Federated Learning (PFL) has emerged as a critical research frontier addressing data heterogeneity issue across distributed clients. Novel model architectures and collaboration mechanisms are engineered to accommodate statistical disparities while producing client-specific models. Parameter decoupling represents a promising paradigm for maintaining model performance in PFL frameworks. However, the communication efficiency of many existing methods remains suboptimal, sustaining substantial communication burdens that impede practical deployment. To bridge this gap, we propose Federated Learning with Programmed Update and Reduced INformation (FedPURIN), a novel framework that strategically identifies critical parameters for transmission through an integer programming formulation. This mathematically grounded strategy is seamlessly integrated into a sparse aggregation scheme, achieving a significant communication reduction while preserving the efficacy. Comprehensive evaluations on standard image classification benchmarks under varied non-IID conditions demonstrate competitive performance relative to state-of-the-art methods, coupled with quantifiable communication reduction through sparse aggregation. The framework establishes a new paradigm for communication-efficient PFL, particularly advantageous for edge intelligence systems operating with heterogeneous data sources.
Authors: Abdelilah Ganmati, Karim Afdel, Lahcen Koutti
Abstract: We present a practical match-on-card design for face verification in which compact 64/128-bit templates are produced off-card by PCA-ITQ and compared on-card via constant-time Hamming distance. We specify ISO/IEC 7816-4 and 14443-4 command APDUs with fixed-length payloads and decision-only status words (no score leakage), together with a minimal per-identity EEPROM map. Using real binary codes from a CelebA working set (55 identities, 412 images), we (i) derive operating thresholds from ROC/DET, (ii) replay enroll->verify transactions at those thresholds, and (iii) bound end-to-end time by pure link latency plus a small constant on-card budget. Even at the slowest contact rate (9.6 kbps), total verification time is 43.9 ms (64 b) and 52.3 ms (128 b); at 38.4 kbps both are <14 ms. At FAR = 1%, both code lengths reach TPR = 0.836, while 128 b lowers EER relative to 64 b. An optional +6 B helper (targeted symbol-level parity over empirically unstable bits) is latency-negligible. Overall, short binary templates, fixed-payload decision-only APDUs, and constant-time matching satisfy ISO/IEC transport constraints with wide timing margin and align with ISO/IEC 24745 privacy goals. Limitations: single-dataset evaluation and design-level (pre-hardware) timing; we outline AgeDB/CFP-FP and on-card microbenchmarks as next steps.
Authors: Jierui Peng, Yanyan Zhang, Yicheng Duan, Tuo Liang, Vipin Chaudhary, Yu Yin
Abstract: The evaluation of Vision-Language-Action (VLA) agents is hindered by the coarse, end-task success metric that fails to provide precise skill diagnosis or measure robustness to real-world perturbations. This challenge is exacerbated by a fragmented data landscape that impedes reproducible research and the development of generalist models. To address these limitations, we introduce \textbf{NEBULA}, a unified ecosystem for single-arm manipulation that enables diagnostic and reproducible evaluation. NEBULA features a novel dual-axis evaluation protocol that combines fine-grained \textit{capability tests} for precise skill diagnosis with systematic \textit{stress tests} that measure robustness. A standardized API and a large-scale, aggregated dataset are provided to reduce fragmentation and support cross-dataset training and fair comparison. Using NEBULA, we demonstrate that top-performing VLAs struggle with key capabilities such as spatial reasoning and dynamic adaptation, which are consistently obscured by conventional end-task success metrics. By measuring both what an agent can do and when it does so reliably, NEBULA provides a practical foundation for robust, general-purpose embodied agents.
Authors: Olajumoke O. Adekunle, Joseph D. Akinyemi, Khadijat T. Ladoja, Olufade F. W. Onifade
Abstract: Lung cancer, a malignancy originating in lung tissues, is commonly diagnosed and classified using medical imaging techniques, particularly computed tomography (CT). Despite the integration of machine learning and deep learning methods, the predictive efficacy of automated systems for lung cancer classification from CT images remains below the desired threshold for clinical adoption. Existing research predominantly focuses on binary classification, distinguishing between malignant and benign lung nodules. In this study, a novel deep learning-based approach is introduced, aimed at an improved multi-class classification, discerning various subtypes of lung cancer from CT images. Leveraging a pre-trained ResNet model, lung tissue images were classified into three distinct classes, two of which denote malignancy and one benign. Employing a dataset comprising 15,000 lung CT images sourced from the LC25000 histopathological images, the ResNet50 model was trained on 10,200 images, validated on 2,550 images, and tested on the remaining 2,250 images. Through the incorporation of custom layers atop the ResNet architecture and meticulous hyperparameter fine-tuning, a remarkable test accuracy of 98.8% was recorded. This represents a notable enhancement over the performance of prior models on the same dataset.
Authors: Junno Yun, Ya\c{s}ar Utku Al\c{c}alar, Mehmet Ak\c{c}akaya
Abstract: Algorithm unrolling methods have proven powerful for solving the regularized least squares problem in computational magnetic resonance imaging (MRI). These approaches unfold an iterative algorithm with a fixed number of iterations, typically alternating between a neural network-based proximal operator for regularization, a data fidelity operation and auxiliary updates with learnable parameters. While the connection to optimization methods dictate that the proximal operator network should be shared across unrolls, this can introduce artifacts or blurring. Heuristically, practitioners have shown that using distinct networks may be beneficial, but this significantly increases the number of learnable parameters, making it challenging to prevent overfitting. To address these shortcomings, by taking inspirations from proximal operators with varying thresholds in approximate message passing (AMP) and the success of time-embedding in diffusion models, we propose a time-embedded algorithm unrolling scheme for inverse problems. Specifically, we introduce a novel perspective on the iteration-dependent proximal operation in vector AMP (VAMP) and the subsequent Onsager correction in the context of algorithm unrolling, framing them as a time-embedded neural network. Similarly, the scalar weights in the data fidelity operation and its associated Onsager correction are cast as time-dependent learnable parameters. Our extensive experiments on the fastMRI dataset, spanning various acceleration rates and datasets, demonstrate that our method effectively reduces aliasing artifacts and mitigates noise amplification, achieving state-of-the-art performance. Furthermore, we show that our time-embedding strategy extends to existing algorithm unrolling approaches, enhancing reconstruction quality without increasing the computational complexity significantly.
Authors: Tong Zhang, Ru Zhang, Jianyi Liu, Zhen Yang, Gongshen Liu
Abstract: Existing concept erasure methods for text-to-image diffusion models commonly rely on fixed anchor strategies, which often lead to critical issues such as concept re-emergence and erosion. To address this, we conduct causal tracing to reveal the inherent sensitivity of erasure to anchor selection and define Sibling Exclusive Concepts as a superior class of anchors. Based on this insight, we propose \textbf{SELECT} (Sibling-Exclusive Evaluation for Contextual Targeting), a dynamic anchor selection framework designed to overcome the limitations of fixed anchors. Our framework introduces a novel two-stage evaluation mechanism that automatically discovers optimal anchors for precise erasure while identifying critical boundary anchors to preserve related concepts. Extensive evaluations demonstrate that SELECT, as a universal anchor solution, not only efficiently adapts to multiple erasure frameworks but also consistently outperforms existing baselines across key performance metrics, averaging only 4 seconds for anchor mining of a single concept.
Authors: Xinfeng Li, Shengyuan Pang, Jialin Wu, Jiangyi Deng, Huanlong Zhong, Yanjiao Chen, Jie Zhang, Wenyuan Xu
Abstract: Text-to-image (T2I) models, though exhibiting remarkable creativity in image generation, can be exploited to produce unsafe images. Existing safety measures, e.g., content moderation or model alignment, fail in the presence of white-box adversaries who know and can adjust model parameters, e.g., by fine-tuning. This paper presents a novel defensive framework, named Patronus, which equips T2I models with holistic protection to defend against white-box adversaries. Specifically, we design an internal moderator that decodes unsafe input features into zero vectors while ensuring the decoding performance of benign input features. Furthermore, we strengthen the model alignment with a carefully designed non-fine-tunable learning mechanism, ensuring the T2I model will not be compromised by malicious fine-tuning. We conduct extensive experiments to validate the intactness of the performance on safe content generation and the effectiveness of rejecting unsafe content generation. Results also confirm the resilience of Patronus against various fine-tuning attacks by white-box adversaries.
Authors: Devin Zhao, Rephael Wenger
Abstract: Let $f: \mathbb{R}^3 \rightarrow \mathbb{R}$ be a scalar field. An isosurface is a piecewise linear approximation of a level set $f^{-1}(\sigma)$ for some $\sigma \in \mathbb{R}$ built from some regular grid sampling of $f$. Isosurfaces constructed from scanned data such as CT scans or MRIs often contain extremely small components that distract from the visualization and do not form part of any geometric model produced from the data. Simple prefiltering of the data can remove such small components while having no effect on the large components that form the body of the visualization. We present experimental results on such filtering.
Authors: Siyin Wang, Wenyi Yu, Xianzhao Chen, Xiaohai Tian, Jun Zhang, Lu Lu, Chao Zhang
Abstract: Human interaction is inherently multimodal and full-duplex: we listen while watching, speak while acting, and fluidly adapt to turn-taking and interruptions. Realizing these capabilities is essential for building models simulating humans. We present ELLSA (End-to-end Listen, Look, Speak and Act), which, to our knowledge, is the first full-duplex, end-to-end model that simultaneously perceives and generates across vision, text, speech, and action within a single architecture, enabling interaction patterns previously out of reach, yielding more natural, human-like behaviors. At its core is a novel SA-MoE architecture (Self-Attention Mixture-of-Experts) that routes each modality to specialized experts and fuses them through a unified attention backbone. This provides a generalizable solution for joint multimodal perception and concurrent generation, leveraging strong pre-trained components while enabling efficient modality integration and mitigating modality interference. On speech-interaction and robot-manipulation benchmarks, ELLSA matches modality-specific baselines, while uniquely supporting advanced multimodal and full-duplex behaviors such as dialogue and action turn-taking, defective instruction rejection, speaking-while-acting, context-grounded visual question answering, and action barge-ins. We contend that ELLSA represents a step toward more natural and general interactive intelligence, contributing to the broader pursuit of artificial general intelligence. All data, code and model checkpoints will be released upon acceptance.
Authors: Simon Jaxy, Anton Theys, Patrick Willett, W. Chris Carleton, Ralf Vandam, Pieter Libin
Abstract: Archaeological predictive modelling estimates where undiscovered sites are likely to occur by combining known locations with environmental, cultural, and geospatial variables. We address this challenge using a deep learning approach but must contend with structural label scarcity inherent to archaeology: positives are rare, and most locations are unlabeled. To address this, we adopt a semi-supervised, positive-unlabeled (PU) learning strategy, implemented as a semantic segmentation model and evaluated on two datasets covering a representative range of archaeological periods. Our approach employs dynamic pseudolabeling, refined with a Conditional Random Field (CRF) implemented via an RNN, increasing label confidence under severe class imbalance. On a geospatial dataset derived from a digital elevation model (DEM), our model performs on par with the state-of-the-art, LAMAP, while achieving higher Dice scores. On raw satellite imagery, assessed end-to-end with stratified k-fold cross-validation, it maintains performance and yields predictive surfaces with improved interpretability. Overall, our results indicate that semi-supervised learning offers a promising approach to identifying undiscovered sites across large, sparsely annotated landscapes.
Authors: Heming Zou, Yunliang Zang, Wutong Xu, Xiangyang Ji
Abstract: Using a nearly-frozen pretrained model, the continual representation learning paradigm reframes parameter updates as a similarity-matching problem to mitigate catastrophic forgetting. However, directly leveraging pretrained features for downstream tasks often suffers from multicollinearity in the similarity-matching stage, and more advanced methods can be computationally prohibitive for real-time, low-latency applications. Inspired by the fly olfactory circuit, we propose Fly-CL, a bio-inspired framework compatible with a wide range of pretrained backbones. Fly-CL substantially reduces training time while achieving performance comparable to or exceeding that of current state-of-the-art methods. We theoretically show how Fly-CL progressively resolves multicollinearity, enabling more effective similarity matching with low time complexity. Extensive simulation experiments across diverse network architectures and data regimes validate Fly-CL's effectiveness in addressing this challenge through a biologically inspired design. Code is available at https://github.com/gfyddha/Fly-CL.
Authors: Hongwei Yan, Guanglong Sun, Zhiqi Kang, Yi Zhong, Liyuan Wang
Abstract: To adapt effectively to dynamic real-world environments, intelligent systems must continually acquire new skills while generalizing them to diverse, unseen scenarios. Here, we introduce a novel and realistic setting named domain generalizable continual learning (DGCL): a model learns sequential tasks with each involving a single domain, aiming to perform well across all encountered tasks and domains. This setting poses unique challenges in acquiring, retaining, and leveraging both semantic- and domain-relevant information for robust generalization. Although state-of-the-art continual learning (CL) methods have employed pre-trained models (PTMs) to enhance task-specific generalization, they typically assume identical training and testing domains for each task and therefore perform poorly in DGCL. To this end, we propose adaptive Domain Transformation (DoT), an innovative PTMs-based approach tailored to DGCL. Inspired by the distributed-plus-hub theory of the human brain, DoT disentangles semantic- and domain-relevant information in representation learning, and adaptively transforms task representations across various domains for output alignment, ensuring balanced and generalized predictions. DoT serves as a plug-in strategy that greatly facilitates state-of-the-art CL baselines under both full parameter tuning and parameter-efficient tuning paradigms in DGCL, validated by extensive experiments. Also, DoT is shown to accumulate domain-generalizable knowledge from DGCL, and ensure resource efficiency with a lightweight implementation.
Authors: Ruiming Guo, Ayush Bhandari
Abstract: The recovery of Dirac impulses, or spikes, from filtered measurements is a classical problem in signal processing. As the spikes lie in the continuous domain while measurements are discrete, this task is known as super-resolution or off-the-grid sparse recovery. Despite significant theoretical and algorithmic advances over the past decade, these developments often overlook critical challenges at the analog-digital interface. In particular, when spikes exhibit strong-weak amplitude disparity, conventional digital acquisition may result in clipping of strong components or loss of weak ones beneath the quantization noise floor. This motivates a broader perspective: super-resolution must simultaneously resolve both amplitude and temporal structure. Under a fixed bit budget, such information loss is unavoidable. In contrast, the emerging theory and practice of the Unlimited Sensing Framework (USF) demonstrate that these fundamental limitations can be overcome. Building on this foundation, we demonstrate that modulo encoding within USF enables digital super-resolution by enhancing measurement precision, thereby unlocking temporal super-resolution beyond conventional limits. We develop new theoretical results that extend to non-bandlimited kernels commonly encountered in practice and introduce a robust algorithm for off-the-grid sparse recovery. To demonstrate practical impact, we instantiate our framework in the context of time-of-flight imaging. Both numerical simulations and hardware experiments validate the effectiveness of our approach under low-bit quantization, enabling super-resolution in amplitude and time.
Authors: Pedram Fekri, Majid Roshanfar, Samuel Barbeau, Seyedfarzad Famouri, Thomas Looi, Dale Podolsky, Mehrdad Zadeh, Javad Dargahi
Abstract: Cardiac catheterization remains a cornerstone of minimally invasive interventions, yet it continues to rely heavily on manual operation. Despite advances in robotic platforms, existing systems are predominantly follow-leader in nature, requiring continuous physician input and lacking intelligent autonomy. This dependency contributes to operator fatigue, more radiation exposure, and variability in procedural outcomes. This work moves towards autonomous catheter navigation by introducing DINO-CVA, a multimodal goal-conditioned behavior cloning framework. The proposed model fuses visual observations and joystick kinematics into a joint embedding space, enabling policies that are both vision-aware and kinematic-aware. Actions are predicted autoregressively from expert demonstrations, with goal conditioning guiding navigation toward specified destinations. A robotic experimental setup with a synthetic vascular phantom was designed to collect multimodal datasets and evaluate performance. Results show that DINO-CVA achieves high accuracy in predicting actions, matching the performance of a kinematics-only baseline while additionally grounding predictions in the anatomical environment. These findings establish the feasibility of multimodal, goal-conditioned architectures for catheter navigation, representing an important step toward reducing operator dependency and improving the reliability of catheterbased therapies.
Authors: Lu Yin, Ziying Shi, Yinghao Wu, Xinyu Yi, Feng Xu, Shihui Guo
Abstract: Human motion capture with sparse inertial sensors has gained significant attention recently. However, existing methods almost exclusively rely on a template adult body shape to model the training data, which poses challenges when generalizing to individuals with largely different body shapes (such as a child). This is primarily due to the variation in IMU-measured acceleration caused by changes in body shape. To fill this gap, we propose Shape-aware Inertial Poser (SAIP), the first solution considering body shape differences in sparse inertial-based motion capture. Specifically, we decompose the sensor measurements related to shape and pose in order to effectively model their joint correlations. Firstly, we train a regression model to transfer the IMU-measured accelerations of a real body to match the template adult body model, compensating for the shape-related sensor measurements. Then, we can easily follow the state-of-the-art methods to estimate the full body motions of the template-shaped body. Finally, we utilize a second regression model to map the joint velocities back to the real body, combined with a shape-aware physical optimization strategy to calculate global motions on the subject. Furthermore, our method relies on body shape awareness, introducing the first inertial shape estimation scheme. This is accomplished by modeling the shape-conditioned IMU-pose correlation using an MLP-based network. To validate the effectiveness of SAIP, we also present the first IMU motion capture dataset containing individuals of different body sizes. This dataset features 10 children and 10 adults, with heights ranging from 110 cm to 190 cm, and a total of 400 minutes of paired IMU-Motion samples. Extensive experimental results demonstrate that SAIP can effectively handle motion capture tasks for diverse body shapes. The code and dataset are available at https://github.com/yinlu5942/SAIP.
Authors: Rishi Sonthalia, Raj Rao Nadakuditi
Abstract: We introduce a novel regularization scheme for autoencoders based on matricial free energy. Our approach defines a differentiable loss function in terms of the singular values of the code matrix (code dimension x batch size). From the standpoint of free probability an d random matrix theory, this loss achieves its minimum when the singular value distribution of the code matrix coincides with that of an appropriately sculpted random metric with i.i.d. Gaussian entries. Empirical simulations demonstrate that minimizing the negative matricial free energy through standard stochastic gradient-based training yields Gaussian-like codes that generalize across training and test sets. Building on this foundation, we propose a matricidal free energy maximizing autoencoder that reliably produces Gaussian codes and show its application to underdetermined inverse problems.
Authors: Yu Gao, Yiru Wang, Anqing Jiang, Heng Yuwen, Wang Shuo, Sun Hao, Wang Jijun
Abstract: Conventional end-to-end (E2E) driving models are effective at generating physically plausible trajectories, but often fail to generalize to long-tail scenarios due to the lack of essential world knowledge to understand and reason about surrounding environments. In contrast, Vision-Language-Action (VLA) models leverage world knowledge to handle challenging cases, but their limited 3D reasoning capability can lead to physically infeasible actions. In this work we introduce DiffVLA++, an enhanced autonomous driving framework that explicitly bridges cognitive reasoning and E2E planning through metric-guided alignment. First, we build a VLA module directly generating semantically grounded driving trajectories. Second, we design an E2E module with a dense trajectory vocabulary that ensures physical feasibility. Third, and most critically, we introduce a metric-guided trajectory scorer that guides and aligns the outputs of the VLA and E2E modules, thereby integrating their complementary strengths. The experiment on the ICCV 2025 Autonomous Grand Challenge leaderboard shows that DiffVLA++ achieves EPDMS of 49.12.
Authors: Yuyang Hong, Qi Yang, Tao Zhang, Zili Wang, Zhaojin Fu, Kun Ding, Bin Fan, Shiming Xiang
Abstract: Recently, significant progress has been made in multi-modal continual learning, aiming to learn new tasks sequentially in multi-modal settings while preserving performance on previously learned ones. However, existing methods mainly focus on coarse-grained tasks, with limitations in addressing modality entanglement in fine-grained continual learning settings. To bridge this gap, we introduce a novel Continual Audio-Visual Segmentation (CAVS) task, aiming to continuously segment new classes guided by audio. Through comprehensive analysis, two critical challenges are identified: 1) multi-modal semantic drift, where a sounding objects is labeled as background in sequential tasks; 2) co-occurrence confusion, where frequent co-occurring classes tend to be confused. In this work, a Collision-based Multi-modal Rehearsal (CMR) framework is designed to address these challenges. Specifically, for multi-modal semantic drift, a Multi-modal Sample Selection (MSS) strategy is proposed to select samples with high modal consistency for rehearsal. Meanwhile, for co-occurence confusion, a Collision-based Sample Rehearsal (CSR) mechanism is designed, allowing for the increase of rehearsal sample frequency of those confusable classes during training process. Moreover, we construct three audio-visual incremental scenarios to verify effectiveness of our method. Comprehensive experiments demonstrate that our method significantly outperforms single-modal continual learning methods.
Authors: Zefan Cai, Haoyi Qiu, Haozhe Zhao, Ke Wan, Jiachen Li, Jiuxiang Gu, Wen Xiao, Nanyun Peng, Junjie Hu
Abstract: Recent advances in video diffusion models have significantly enhanced text-to-video generation, particularly through alignment tuning using reward models trained on human preferences. While these methods improve visual quality, they can unintentionally encode and amplify social biases. To systematically trace how such biases evolve throughout the alignment pipeline, we introduce VideoBiasEval, a comprehensive diagnostic framework for evaluating social representation in video generation. Grounded in established social bias taxonomies, VideoBiasEval employs an event-based prompting strategy to disentangle semantic content (actions and contexts) from actor attributes (gender and ethnicity). It further introduces multi-granular metrics to evaluate (1) overall ethnicity bias, (2) gender bias conditioned on ethnicity, (3) distributional shifts in social attributes across model variants, and (4) the temporal persistence of bias within videos. Using this framework, we conduct the first end-to-end analysis connecting biases in human preference datasets, their amplification in reward models, and their propagation through alignment-tuned video diffusion models. Our results reveal that alignment tuning not only strengthens representational biases but also makes them temporally stable, producing smoother yet more stereotyped portrayals. These findings highlight the need for bias-aware evaluation and mitigation throughout the alignment process to ensure fair and socially responsible video generation.
Authors: Ludovica Schaerf
Abstract: This paper examines the evolving nature of internal representations in generative visual models, focusing on the conceptual and technical shift from GANs and VAEs to diffusion-based architectures. Drawing on Beatrice Fazi's account of synthesis as the amalgamation of distributed representations, we propose a distinction between "synthesis in a strict sense", where a compact latent space wholly determines the generative process, and "synthesis in a broad sense," which characterizes models whose representational labor is distributed across layers. Through close readings of model architectures and a targeted experimental setup that intervenes in layerwise representations, we show how diffusion models fragment the burden of representation and thereby challenge assumptions of unified internal space. By situating these findings within media theoretical frameworks and critically engaging with metaphors such as the latent space and the Platonic Representation Hypothesis, we argue for a reorientation of how generative AI is understood: not as a direct synthesis of content, but as an emergent configuration of specialized processes.
Authors: Alejandro Guerra-Manzanares, Farah E. Shamout
Abstract: The aim of multimodal neural networks is to combine diverse data sources, referred to as modalities, to achieve enhanced performance compared to relying on a single modality. However, training of multimodal networks is typically hindered by modality overfitting, where the network relies excessively on one of the available modalities. This often yields sub-optimal performance, hindering the potential of multimodal learning and resulting in marginal improvements relative to unimodal models. In this work, we present the Modality-Informed Learning ratE Scheduler (MILES) for training multimodal joint fusion models in a balanced manner. MILES leverages the differences in modality-wise conditional utilization rates during training to effectively balance multimodal learning. The learning rate is dynamically adjusted during training to balance the speed of learning from each modality by the multimodal model, aiming for enhanced performance in both multimodal and unimodal predictions. We extensively evaluate MILES on four multimodal joint fusion tasks and compare its performance to seven state-of-the-art baselines. Our results show that MILES outperforms all baselines across all tasks and fusion methods considered in our study, effectively balancing modality usage during training. This results in improved multimodal performance and stronger modality encoders, which can be leveraged when dealing with unimodal samples or absent modalities. Overall, our work highlights the impact of balancing multimodal learning on improving model performance.
Authors: Zhengshen Zhang, Hao Li, Yalun Dai, Zhengbang Zhu, Lei Zhou, Chenchen Liu, Dong Wang, Francis E. H. Tay, Sijin Chen, Ziwei Liu, Yuxiao Liu, Xinghang Li, Pan Zhou
Abstract: Existing vision-language-action (VLA) models act in 3D real-world but are typically built on 2D encoders, leaving a spatial reasoning gap that limits generalization and adaptability. Recent 3D integration techniques for VLAs either require specialized sensors and transfer poorly across modalities, or inject weak cues that lack geometry and degrade vision-language alignment. In this work, we introduce FALCON (From Spatial to Action), a novel paradigm that injects rich 3D spatial tokens into the action head. FALCON leverages spatial foundation models to deliver strong geometric priors from RGB alone, and includes an Embodied Spatial Model that can optionally fuse depth, or pose for higher fidelity when available, without retraining or architectural changes. To preserve language reasoning, spatial tokens are consumed by a Spatial-Enhanced Action Head rather than being concatenated into the vision-language backbone. These designs enable FALCON to address limitations in spatial representation, modality transferability, and alignment. In comprehensive evaluations across three simulation benchmarks and eleven real-world tasks, our proposed FALCON achieves state-of-the-art performance, consistently surpasses competitive baselines, and remains robust under clutter, spatial-prompt conditioning, and variations in object scale and height.
Authors: Olivier Parisot, Mahmoud Jaziri
Abstract: The growing negative impact of the visibility of satellites in the night sky is influencing the practice of astronomy and astrophotograph, both at the amateur and professional levels. The presence of these satellites has the effect of introducing streaks into the images captured during astronomical observation, requiring the application of additional post processing to mitigate the undesirable impact, whether for data loss or cosmetic reasons. In this paper, we show how we test and adapt various Deep Learning approaches to detect streaks in raw astronomical data captured between March 2022 and February 2023 with smart telescopes.
Authors: Mir Nafis Sharear Shopnil, Sharad Duwal, Abhishek Tyagi, Adiba Mahbub Proma
Abstract: Misinformation spreads across web platforms through billions of daily multimodal posts that combine text and images, overwhelming manual fact-checking capacity. Supervised detection models require domain-specific training data and fail to generalize across diverse manipulation tactics. We present MIRAGE, an inference-time, model-pluggable agentic framework that decomposes multimodal verification into four sequential modules: visual veracity assessment detects AI-generated images, cross-modal consistency analysis identifies out-of-context repurposing, retrieval-augmented factual checking grounds claims in web evidence through iterative question generation, and a calibrated judgment module integrates all signals. MIRAGE orchestrates vision-language model reasoning with targeted web retrieval, outputs structured and citation-linked rationales. On MMFakeBench validation set (1,000 samples), MIRAGE with GPT-4o-mini achieves 81.65% F1 and 75.1% accuracy, outperforming the strongest zero-shot baseline (GPT-4V with MMD-Agent at 74.0% F1) by 7.65 points while maintaining 34.3% false positive rate versus 97.3% for a judge-only baseline. Test set results (5,000 samples) confirm generalization with 81.44% F1 and 75.08% accuracy. Ablation studies show visual verification contributes 5.18 F1 points and retrieval-augmented reasoning contributes 2.97 points. Our results demonstrate that decomposed agentic reasoning with web retrieval can match supervised detector performance without domain-specific training, enabling misinformation detection across modalities where labeled data remains scarce.
Authors: Hendric Voss, Lisa Michelle Bohnenkamp, Stefan Kopp
Abstract: This study explores two frameworks for co-speech gesture generation, AQ-GT and its semantically-augmented variant AQ-GT-a, to evaluate their ability to convey meaning through gestures and how humans perceive the resulting movements. Using sentences from the SAGA spatial communication corpus, contextually similar sentences, and novel movement-focused sentences, we conducted a user-centered evaluation of concept recognition and human-likeness. Results revealed a nuanced relationship between semantic annotations and performance. The original AQ-GT framework, lacking explicit semantic input, was surprisingly more effective at conveying concepts within its training domain. Conversely, the AQ-GT-a framework demonstrated better generalization, particularly for representing shape and size in novel contexts. While participants rated gestures from AQ-GT-a as more expressive and helpful, they did not perceive them as more human-like. These findings suggest that explicit semantic enrichment does not guarantee improved gesture generation and that its effectiveness is highly dependent on the context, indicating a potential trade-off between specialization and generalization.
Authors: Hendric Voss, Stefan Kopp
Abstract: Human communication combines speech with expressive nonverbal cues such as hand gestures that serve manifold communicative functions. Yet, current generative gesture generation approaches are restricted to simple, repetitive beat gestures that accompany the rhythm of speaking but do not contribute to communicating semantic meaning. This paper tackles a core challenge in co-speech gesture synthesis: generating iconic or deictic gestures that are semantically coherent with a verbal utterance. Such gestures cannot be derived from language input alone, which inherently lacks the visual meaning that is often carried autonomously by gestures. We therefore introduce a zero-shot system that generates gestures from a given language input and additionally is informed by imagistic input, without manual annotation or human intervention. Our method integrates an image analysis pipeline that extracts key object properties such as shape, symmetry, and alignment, together with a semantic matching module that links these visual details to spoken text. An inverse kinematics engine then synthesizes iconic and deictic gestures and combines them with co-generated natural beat gestures for coherent multimodal communication. A comprehensive user study demonstrates the effectiveness of our approach. In scenarios where speech alone was ambiguous, gestures generated by our system significantly improved participants' ability to identify object properties, confirming their interpretability and communicative value. While challenges remain in representing complex shapes, our results highlight the importance of context-aware semantic gestures for creating expressive and collaborative virtual agents or avatars, marking a substantial step forward towards efficient and robust, embodied human-agent interaction. More information and example videos are available here: https://review-anon-io.github.io/ImaGGen.github.io/
Authors: Athanasios Angelakis, Amne Mousa, Micah L. A. Heldeweg, Laurens A. Biesheuvel, Mark A. Haaksma, Jasper M. Smit, Pieter R. Tuinman, Paul W. G. Elbers
Abstract: Differentiating cardiogenic pulmonary oedema (CPE) from non-cardiogenic and structurally normal lungs in lung ultrasound (LUS) videos remains challenging due to the high visual variability of non-cardiogenic inflammatory patterns (NCIP/ARDS-like), interstitial lung disease, and healthy lungs. This heterogeneity complicates automated classification as overlapping B-lines and pleural artefacts are common. We introduce ZACH-ViT (Zero-token Adaptive Compact Hierarchical Vision Transformer), a 0.25 M-parameter Vision Transformer variant that removes both positional embeddings and the [CLS] token, making it fully permutation-invariant and suitable for unordered medical image data. To enhance generalization, we propose ShuffleStrides Data Augmentation (SSDA), which permutes probe-view sequences and frame orders while preserving anatomical validity. ZACH-ViT was evaluated on 380 LUS videos from 95 critically ill patients against nine state-of-the-art baselines. Despite the heterogeneity of the non-cardiogenic group, ZACH-ViT achieved the highest validation and test ROC-AUC (0.80 and 0.79) with balanced sensitivity (0.60) and specificity (0.91), while all competing models collapsed to trivial classification. It trains 1.35x faster than Minimal ViT (0.62M parameters) with 2.5x fewer parameters, supporting real-time clinical deployment. These results show that aligning architectural design with data structure can outperform scale in small-data medical imaging.
Authors: Qilin Liao, Anamika Lochab, Ruqi Zhang
Abstract: Vision-Language Models (VLMs) extend large language models with visual reasoning, but their multimodal design also introduces new, underexplored vulnerabilities. Existing multimodal red-teaming methods largely rely on brittle templates, focus on single-attack settings, and expose only a narrow subset of vulnerabilities. To address these limitations, we introduce VERA-V, a variational inference framework that recasts multimodal jailbreak discovery as learning a joint posterior distribution over paired text-image prompts. This probabilistic view enables the generation of stealthy, coupled adversarial inputs that bypass model guardrails. We train a lightweight attacker to approximate the posterior, allowing efficient sampling of diverse jailbreaks and providing distributional insights into vulnerabilities. VERA-V further integrates three complementary strategies: (i) typography-based text prompts that embed harmful cues, (ii) diffusion-based image synthesis that introduces adversarial signals, and (iii) structured distractors to fragment VLM attention. Experiments on HarmBench and HADES benchmarks show that VERA-V consistently outperforms state-of-the-art baselines on both open-source and frontier VLMs, achieving up to 53.75% higher attack success rate (ASR) over the best baseline on GPT-4o.
Authors: Zhining Liu, Ziyi Chen, Hui Liu, Chen Luo, Xianfeng Tang, Suhang Wang, Joy Zeng, Zhenwei Dai, Zhan Shi, Tianxin Wei, Benoit Dumoulin, Hanghang Tong
Abstract: Vision-Language Models (VLMs) achieve strong results on multimodal tasks such as visual question answering, yet they can still fail even when the correct visual evidence is present. In this work, we systematically investigate whether these failures arise from not perceiving the evidence or from not leveraging it effectively. By examining layer-wise attention dynamics, we find that shallow layers focus primarily on text, while deeper layers sparsely but reliably attend to localized evidence regions. Surprisingly, VLMs often perceive the visual evidence when outputting incorrect answers, a phenomenon we term ``seeing but not believing'' that widely exists in major VLM families. Building on this, we introduce an inference-time intervention that highlights deep-layer evidence regions through selective attention-based masking. It requires no training and consistently improves accuracy across multiple families, including LLaVA, Qwen, Gemma, and InternVL. These results show that VLMs encode reliable evidence internally but under-utilize it, making such signals explicit can bridge the gap between perception and reasoning, advancing the diagnostic understanding and reliability of VLMs.
Authors: Simeon Adebola, Chung Min Kim, Justin Kerr, Shuangyu Xie, Prithvi Akella, Jose Luis Susa Rincon, Eugen Solowjow, Ken Goldberg
Abstract: Commercial plant phenotyping systems using fixed cameras cannot perceive many plant details due to leaf occlusion. In this paper, we present Botany-Bot, a system for building detailed "annotated digital twins" of living plants using two stereo cameras, a digital turntable inside a lightbox, an industrial robot arm, and 3D segmentated Gaussian Splat models. We also present robot algorithms for manipulating leaves to take high-resolution indexable images of occluded details such as stem buds and the underside/topside of leaves. Results from experiments suggest that Botany-Bot can segment leaves with 90.8% accuracy, detect leaves with 86.2% accuracy, lift/push leaves with 77.9% accuracy, and take detailed overside/underside images with 77.3% accuracy. Code, videos, and datasets are available at https://berkeleyautomation.github.io/Botany-Bot/.
Authors: Kaicheng Pang, Xingxing Zou, Waikeung Wong
Abstract: Fashion compatibility models enable online retailers to easily obtain a large number of outfit compositions with good quality. However, effective fashion recommendation demands precise service for each customer with a deeper cognition of fashion. In this paper, we conduct the first study on fashion cognitive learning, which is fashion recommendations conditioned on personal physical information. To this end, we propose a Fashion Cognitive Network (FCN) to learn the relationships among visual-semantic embedding of outfit composition and appearance features of individuals. FCN contains two submodules, namely outfit encoder and Multi-label Graph Neural Network (ML-GCN). The outfit encoder uses a convolutional layer to encode an outfit into an outfit embedding. The latter module learns label classifiers via stacked GCN. We conducted extensive experiments on the newly collected O4U dataset, and the results provide strong qualitative and quantitative evidence that our framework outperforms alternative methods.
Authors: Junho Kim, Young Min Kim, Ramzi Zahreddine, Weston A. Welge, Gurunandan Krishnan, Sizhuo Ma, Jian Wang
Abstract: We consider the problem of client-server localization, where edge device users communicate visual data with the service provider for locating oneself against a pre-built 3D map. This localization paradigm is a crucial component for location-based services in AR/VR or mobile applications, as it is not trivial to store large-scale 3D maps and process fast localization on resource-limited edge devices. Nevertheless, conventional client-server localization systems possess numerous challenges in computational efficiency, robustness, and privacy-preservation during data transmission. Our work aims to jointly solve these challenges with a localization pipeline based on event cameras. By using event cameras, our system consumes low energy and maintains small memory bandwidth. Then during localization, we propose applying event-to-image conversion and leverage mature image-based localization, which achieves robustness even in low-light or fast-moving scenes. To further enhance privacy protection, we introduce privacy protection techniques at two levels. Network level protection aims to hide the entire user's view in private scenes using a novel split inference approach, while sensor level protection aims to hide sensitive user details such as faces with light-weight filtering. Both methods involve small client-side computation and localization performance loss, while significantly mitigating the feeling of insecurity as revealed in our user study. We thus project our method to serve as a building block for practical location-based services using event cameras. Project page including the code is available through this link: https://82magnolia.github.io/event\_localization/.
Authors: Qiang Fu, Matheus Souza, Eunsue Choi, Suhyun Shin, Seung-Hwan Baek, Wolfgang Heidrich
Abstract: Hyperspectral imaging empowers machine vision systems with the distinct capability of identifying materials through recording their spectral signatures. Recent efforts in data-driven spectral reconstruction aim at extracting spectral information from RGB images captured by cost-effective RGB cameras, instead of dedicated hardware. Published work reports exceedingly high numerical scores for this reconstruction task, yet real-world performance lags substantially behind. We systematically analyze the performance of such methods. First, we evaluate the overfitting limitations with respect to current datasets by training the networks with less data, validating the trained models with unseen yet slightly modified data and cross-dataset validation. Second, we reveal fundamental limitations in the ability of RGB to spectral methods to deal with metameric or near-metameric conditions, which have so far gone largely unnoticed due to the insufficiencies of existing datasets. We validate the trained models with metamer data generated by metameric black theory and re-training the networks with various forms of metamers. This methodology can also be used for data augmentation as a partial mitigation of the dataset issues, although the RGB to spectral inverse problem remains fundamentally ill-posed. Finally, we analyze the potential for modifying the problem setting to achieve better performance by exploiting optical encoding provided by either optical aberrations or deliberate optical design. Our experiments show such approaches provide improved results under certain circumstances, but their overall performance is limited by the same dataset issues. We conclude that future progress on snapshot spectral imaging will heavily depend on the generation of improved datasets which can then be used to design effective optical encoding strategies. Code: https://github.com/vccimaging/OpticsAwareHSI-Analysis.
URLs: https://github.com/vccimaging/OpticsAwareHSI-Analysis.
Authors: Rohit Jena, Pratik Chaudhari, James C. Gee
Abstract: The paper proposes FireANTs, a multi-scale Adaptive Riemannian Optimization algorithm for dense diffeomorphic image matching. Existing state-of-the-art methods for diffeomorphic image matching are slow due to inefficient implementations and slow convergence due to the ill-conditioned nature of the optimization problem. Deep learning methods offer fast inference but require extensive training time, substantial inference memory, and fail to generalize across long-tailed distributions or diverse image modalities, necessitating costly retraining. We address these challenges by proposing a training-free, GPU-accelerated multi-scale Adaptive Riemannian Optimization algorithm for fast and accurate dense diffeomorphic image matching. FireANTs runs about 2.5x faster than ANTs on a CPU, and upto 1200x faster on a GPU. On a single GPU, FireANTs performs competitively with deep learning methods on inference runtime while consuming upto 10x less memory. FireANTs shows remarkable robustness to a wide variety of matching problems across modalities, species, and organs without any domain-specific training or tuning. Our framework allows hyperparameter grid search studies with significantly less resources and time compared to traditional and deep learning registration algorithms alike.
Authors: Xinghan Wang, Zixi Kang, Yadong Mu
Abstract: Human motion understanding is a fundamental task with diverse practical applications, facilitated by the availability of large-scale motion capture datasets. Recent studies focus on text-motion tasks, such as text-based motion generation, editing and question answering. In this study, we introduce the novel task of text-based human motion grounding (THMG), aimed at precisely localizing temporal segments corresponding to given textual descriptions within untrimmed motion sequences. Capturing global temporal information is crucial for the THMG task. However, Transformer-based models that rely on global temporal self-attention face challenges when handling long untrimmed sequences due to the quadratic computational cost. We address these challenges by proposing Text-controlled Motion Mamba (TM-Mamba), a unified model that integrates temporal global context, language query control, and spatial graph topology with only linear memory cost. The core of the model is a text-controlled selection mechanism which dynamically incorporates global temporal information based on text query. The model is further enhanced to be topology-aware through the integration of relational embeddings. For evaluation, we introduce BABEL-Grounding, the first text-motion dataset that provides detailed textual descriptions of human actions along with their corresponding temporal segments. Extensive evaluations demonstrate the effectiveness of TM-Mamba on BABEL-Grounding.
Authors: Mykhailo Uss, Ruslan Yermolenko, Oleksii Shashko, Olena Kolodiazhna, Ivan Safonov, Volodymyr Savin, Yoonjae Yeo, Seowon Ji, Jaeyun Jeong
Abstract: Dense depth prediction deep neural networks (DNN) have achieved impressive results for both monocular and binocular data, but still they are limited by high computational complexity, restricting their use on low-end devices. For better on-device efficiency and hardware utilization, weights and activations of the DNN should be converted to low-bit precision. However, this precision is not sufficient to represent high dynamic range depth. In this paper, we aim to overcome this limitation and restore high-precision depth from low-bit precision predictions. To achieve this, we propose to represent high dynamic range depth as two low dynamic range components of a Hilbert curve, and to train the full-precision DNN to directly predict the latter. For on-device deployment, we use standard quantization methods and add a post-processing step that reconstructs depth from the Hilbert curve components predicted in low-bit precision. Extensive experiments demonstrate that our method increases the bit precision of predicted depth by up to three bits with little computational overhead. We also observed a positive side effect of quantization error reduction by up to 4.6 times. Our method enables effective and accurate depth prediction with DNN weights and activations quantized to eight-bit precision.
Authors: Sooyeon Go, Kyungmook Choi, Minjung Shin, Youngjung Uh
Abstract: As pre-trained text-to-image diffusion models have become a useful tool for image synthesis, people want to specify the results in various ways. This paper tackles training-free appearance transfer, which produces an image with the structure of a target image from the appearance of a reference image. Existing methods usually do not reflect semantic correspondence, as they rely on query-key similarity within the self-attention layer to establish correspondences between images. To this end, we propose explicitly rearranging the features according to the dense semantic correspondences. Extensive experiments show the superiority of our method in various aspects: preserving the structure of the target and reflecting the correct color from the reference, even when the two images are not aligned.
Authors: Ling Li, Yu Ye, Yao Zhou, Wei Zeng
Abstract: This work tackles the problem of geo-localization with a new paradigm using a large vision-language model (LVLM) augmented with human inference knowledge. A primary challenge here is the scarcity of data for training the LVLM - existing street-view datasets often contain numerous low-quality images lacking visual clues, and lack any reasoning inference. To address the data-quality issue, we devise a CLIP-based network to quantify the degree of street-view images being locatable, leading to the creation of a new dataset comprising highly locatable street views. To enhance reasoning inference, we integrate external knowledge obtained from real geo-localization games, tapping into valuable human inference capabilities. The data are utilized to train GeoReasoner, which undergoes fine-tuning through dedicated reasoning and location-tuning stages. Qualitative and quantitative evaluations illustrate that GeoReasoner outperforms counterpart LVLMs by more than 25% at country-level and 38% at city-level geo-localization tasks, and surpasses StreetCLIP performance while requiring fewer training resources. The data and code are available at https://github.com/lingli1996/GeoReasoner.
Authors: Siqi Luo, Yi Xin, Yuntao Du, Zhongwei Wan, Tao Tan, Guangtao Zhai, Xiaohong Liu
Abstract: Deep neural networks often encounter significant performance drops while facing with domain shifts between training (source) and test (target) data. To address this issue, Test Time Adaptation (TTA) methods have been proposed to adapt pre-trained source model to handle out-of-distribution streaming target data. Although these methods offer some relief, they lack a reliable mechanism for domain shift correction, which can often be erratic in real-world applications. In response, we develop Few-Shot Test Time Adaptation (FS-TTA), a novel and practical setting that utilizes a few-shot support set on top of TTA. Adhering to the principle of few inputs, big gains, FS-TTA reduces blind exploration in unseen target domains. Furthermore, we propose a two-stage framework to tackle FS-TTA, including (i) fine-tuning the pre-trained source model with few-shot support set, along with using feature diversity augmentation module to avoid overfitting, (ii) implementing test time adaptation based on prototype memory bank guidance to produce high quality pseudo-label for model adaptation. Through extensive experiments on three cross-domain classification benchmarks, we demonstrate the superior performance and reliability of our FS-TTA and framework.
Authors: Junho Lee, Jeongwoo Shin, Seung Woo Ko, Seongsu Ha, Joonseok Lee
Abstract: Given a video with $T$ frames, frame sampling is a task to select $N \ll T$ frames, so as to maximize the performance of a fixed video classifier. Not just brute-force search, but most existing methods suffer from its vast search space of $\binom{T}{N}$, especially when $N$ gets large. To address this challenge, we introduce a novel perspective of reducing the search space from $O(T^N)$ to $O(T)$. Instead of exploring the entire $O(T^N)$ space, our proposed semi-optimal policy selects the top $N$ frames based on the independently estimated value of each frame using per-frame confidence, significantly reducing the computational complexity. We verify that our semi-optimal policy can efficiently approximate the optimal policy, particularly under practical settings. Additionally, through extensive experiments on various datasets and model architectures, we demonstrate that learning our semi-optimal policy ensures stable and high performance regardless of the size of $N$ and $T$.
Authors: Hua Yan, Heng Tan, Yi Ding, Pengfei Zhou, Vinod Namboodiri, Yu Yang
Abstract: Human Activity Recognition (HAR) using Inertial Measurement Unit (IMU) sensors is critical for applications in healthcare, safety, and industrial production. However, variations in activity patterns, device types, and sensor placements create distribution gaps across datasets, reducing the performance of HAR models. To address this, we propose LanHAR, a novel system that leverages Large Language Models (LLMs) to generate semantic interpretations of sensor readings and activity labels for cross-dataset HAR. This approach not only mitigates cross-dataset heterogeneity but also enhances the recognition of new activities. LanHAR employs an iterative re-generation method to produce high-quality semantic interpretations with LLMs and a two-stage training framework that bridges the semantic interpretations of sensor readings and activity labels. This ultimately leads to a lightweight sensor encoder suitable for mobile deployment, enabling any sensor reading to be mapped into the semantic interpretation space. Experiments on five public datasets demonstrate that our approach significantly outperforms state-of-the-art methods in both cross-dataset HAR and new activity recognition. The source code is publicly available at https://github.com/DASHLab/LanHAR.
Authors: Nhan T. Luu, Duong T. Luu, Nam N. Pham, Thang C. Truong
Abstract: Spiking neural network (SNN) has emerged as a promising paradigm in computational neuroscience and artificial intelligence, offering advantages such as low energy consumption and small memory footprint. However, their practical adoption is constrained by several challenges, prominently among them being performance optimization. In this study, we present a novel approach to enhance the performance of SNN for images through a new coding method that exploits bit plane representation. Our proposed technique is designed to improve the accuracy of SNN without increasing model size. Also, we investigate the impacts of color models of the proposed coding process. Through extensive experimental validation, we demonstrate the effectiveness of our coding strategy in achieving performance gain across multiple datasets. To the best of our knowledge, this is the first research that considers bit planes and color models in the context of SNN. By leveraging the unique characteristics of bit planes, we hope to unlock new potentials in SNNs performance, potentially paving the way for more efficient and effective SNNs models in future researches and applications.
Authors: Ekkasit Pinyoanuntapong, Muhammad Usama Saleem, Korrawe Karunratanakul, Pu Wang, Hongfei Xue, Chen Chen, Chuan Guo, Junli Cao, Jian Ren, Sergey Tulyakov
Abstract: Recent advances in motion diffusion models have enabled spatially controllable text-to-motion generation. However, these models struggle to achieve high-precision control while maintaining high-quality motion generation. To address these challenges, we propose MaskControl, the first approach to introduce controllability to the generative masked motion model. Our approach introduces two key innovations. First, \textit{Logits Regularizer} implicitly perturbs logits at training time to align the distribution of motion tokens with the controlled joint positions, while regularizing the categorical token prediction to ensure high-fidelity generation. Second, \textit{Logit Optimization} explicitly optimizes the predicted logits during inference time, directly reshaping the token distribution that forces the generated motion to accurately align with the controlled joint positions. Moreover, we introduce \textit{Differentiable Expectation Sampling (DES)} to combat the non-differential distribution sampling process encountered by logits regularizer and optimization. Extensive experiments demonstrate that MaskControl outperforms state-of-the-art methods, achieving superior motion quality (FID decreases by ~77\%) and higher control precision (average error 0.91 vs. 1.08). Additionally, MaskControl enables diverse applications, including any-joint-any-frame control, body-part timeline control, and zero-shot objective control. Video visualization can be found at https://www.ekkasit.com/ControlMM-page/
Authors: Mehdi Hosseini Chagahi, Saeed Mohammadi Dashtaki, Niloufar Delfan, Nadia Mohammadi, Farshid Rostami Pouria, Behzad Moshiri, Md. Jalil Piran, Oliver Faust
Abstract: Osteoporosis is a common condition that increases fracture risk, especially in older adults. Early diagnosis is vital for preventing fractures, reducing treatment costs, and preserving mobility. However, healthcare providers face challenges like limited labeled data and difficulties in processing medical images. This study presents a novel multi-modal learning framework that integrates clinical and imaging data to improve diagnostic accuracy and model interpretability. The model utilizes three pre-trained networks-VGG19, InceptionV3, and ResNet50-to extract deep features from X-ray images. These features are transformed using PCA to reduce dimensionality and focus on the most relevant components. A clustering-based selection process identifies the most representative components, which are then combined with preprocessed clinical data and processed through a fully connected network (FCN) for final classification. A feature importance plot highlights key variables, showing that Medical History, BMI, and Height were the main contributors, emphasizing the significance of patient-specific data. While imaging features were valuable, they had lower importance, indicating that clinical data are crucial for accurate predictions. This framework promotes precise and interpretable predictions, enhancing transparency and building trust in AI-driven diagnoses for clinical integration.
Authors: Wenjie Li, Jiawei Li, Pengcheng Zeng, Christian Schroeder de Witt, Ameya Prabhu, Amartya Sanyal
Abstract: Addressing data integrity challenges, such as unlearning the effects of data poisoning after model training, is necessary for the reliable deployment of machine learning models. State-of-the-art influence functions, such as EK-FAC and TRAK, often fail to accurately attribute abnormal model behavior to the specific poisoned training data responsible for the data poisoning attack. In addition, traditional unlearning algorithms often struggle to effectively remove the influence of poisoned samples, particularly when only a few affected examples can be identified. To address these challenge, we introduce $\Delta$-Influence, a novel approach that leverages influence functions to trace abnormal model behavior back to the responsible poisoned training data using as little as just one poisoned test example. $\Delta$-Influence applies data transformations that sever the link between poisoned training data and compromised test points without significantly affecting clean data. This allows $\Delta$-Influence to detect large negative shifts in influence scores following data transformations, a phenomenon we term as influence collapse, thereby accurately identifying poisoned training data. Unlearning this subset, e.g. through retraining, effectively eliminates the data poisoning. We validate our method across three vision-based poisoning attacks and three datasets, benchmarking against five detection algorithms and five unlearning strategies. We show that $\Delta$-Influence consistently achieves the best unlearning across all settings, showing the promise of influence functions for corrective unlearning. Our code is publicly available at: https://github.com/Ruby-a07/delta-influence
Authors: Wang Bill Zhu, Deqing Fu, Kai Sun, Yi Lu, Zhaojiang Lin, Seungwhan Moon, Kanika Narang, Mustafa Canim, Yue Liu, Anuj Kumar, Xin Luna Dong
Abstract: Existing recommendation systems either rely on user interaction logs, such as online shopping history for shopping recommendations, or focus on text signals. However, item-based histories are not always accessible, and are not generalizable for multimodal recommendation. We hypothesize that a user's visual history -- comprising images from daily life -- can offer rich, task-agnostic insights into their interests and preferences, and thus be leveraged for effective personalization. To this end, we propose VisualLens, a novel framework that leverages multimodal large language models (MLLMs) to enable personalization using task-agnostic visual history. VisualLens extracts, filters, and refines a spectrum user profile from the visual history to support personalized recommendation. We created two new benchmarks, Google-Review-V and Yelp-V, with task-agnostic visual histories, and show that VisualLens improves over state-of-the-art item-based multimodal recommendations by 5-10% on Hit@3, and outperforms GPT-4o by 2-5%. Further analysis shows that VisualLens is robust across varying history lengths and excels at adapting to both longer histories and unseen content categories.
Authors: Haojie Zhang, Zhihao Liang, Ruibo Fu, Bingyan Liu, Zhengqi Wen, Xuefei Liu, Jianhua Tao, Yaling Liang
Abstract: Long-duration talking video synthesis faces enduring challenges in achieving high video quality, portrait and temporal consistency, and computational efficiency. As video length increases, issues such as visual degradation, identity inconsistency, temporal incoherence, and error accumulation become increasingly problematic, severely affecting the realism and reliability of the results. To address these challenges, we present LetsTalk, a diffusion transformer framework equipped with multimodal guidance and a novel memory bank mechanism, explicitly maintaining contextual continuity and enabling robust, high-quality, and efficient generation of long-duration talking videos. In particular, LetsTalk introduces a noise-regularized memory bank to alleviate error accumulation and sampling artifacts during extended video generation. To further improve efficiency and spatiotemporal consistency, LetsTalk employs a deep compression autoencoder and a spatiotemporal-aware transformer with linear attention for effective multimodal fusion. We systematically analyze three fusion schemes and show that combining deep (Symbiotic Fusion) for portrait features and shallow (Direct Fusion) for audio achieves superior visual realism and precise speech-driven motion, while preserving diversity of movements. Extensive experiments demonstrate that LetsTalk establishes new state-of-the-art in generation quality, producing temporally coherent and realistic talking videos with enhanced diversity and liveliness, and maintains remarkable efficiency with 8x fewer parameters than previous approaches.
Authors: Jaemin Kim, Bryan Sangwoo Kim, Jong Chul Ye
Abstract: Diffusion models have achieved impressive results in generative tasks for text-to-video (T2V) synthesis. However, achieving accurate text alignment in T2V generation remains challenging due to the complex temporal dependencies across frames. Existing reinforcement learning (RL)-based approaches to enhance text alignment often require differentiable reward functions trained for videos, hindering their scalability and applicability. In this paper, we propose \textbf{Free$^2$Guide}, a novel gradient-free and training-free framework for aligning generated videos with text prompts. Specifically, leveraging principles from path integral control, Free$^2$Guide approximates guidance for diffusion models using non-differentiable reward functions, thereby enabling the integration of powerful black-box Large Vision-Language Models (LVLMs) as reward models. To enable image-trained LVLMs to assess text-to-video alignment, we leverage \textit{stitching} between video frames and use system prompts to capture sequential attributions. Our framework supports the flexible ensembling of multiple reward models to synergistically enhance alignment without significant computational overhead. Experimental results confirm that Free$^2$Guide using image-trained LVLMs significantly improves text-to-video alignment, thereby enhancing the overall video quality. Our results and code are available at https://kjm981995.github.io/free2guide/
Authors: Xiaofeng Tan, Hongsong Wang, Xin Geng, Pan Zhou
Abstract: Text-to-motion generation is essential for advancing the creative industry but often presents challenges in producing consistent, realistic motions. To address this, we focus on fine-tuning text-to-motion models to consistently favor high-quality, human-preferred motions, a critical yet largely unexplored problem. In this work, we theoretically investigate the DPO under both online and offline settings, and reveal their respective limitation: overfitting in offline DPO, and biased sampling in online DPO. Building on our theoretical insights, we introduce Semi-online Preference Optimization (SoPo), a DPO-based method for training text-to-motion models using "semi-online" data pair, consisting of unpreferred motion from online distribution and preferred motion in offline datasets. This method leverages both online and offline DPO, allowing each to compensate for the other's limitations. Extensive experiments demonstrate that SoPo outperforms other preference alignment methods, with an MM-Dist of 3.25% (vs e.g. 0.76% of MoDiPO) on the MLD model, 2.91% (vs e.g. 0.66% of MoDiPO) on MDM model, respectively. Additionally, the MLD model fine-tuned by our SoPo surpasses the SoTA model in terms of R-precision and MM Dist. Visualization results also show the efficacy of our SoPo in preference alignment. Project page: https://xiaofeng-tan.github.io/projects/SoPo/ .
Authors: Yilei Jiang, Weihong Li, Yiyuan Zhang, Minghong Cai, Xiangyu Yue
Abstract: While Diffusion Models (DM) exhibit remarkable performance across various image generative tasks, they nonetheless reflect the inherent bias presented in the training set. As DMs are now widely used in real-world applications, these biases could perpetuate a distorted worldview and hinder opportunities for minority groups. Existing methods on debiasing DMs usually requires model retraining with a human-crafted reference dataset or additional classifiers, which suffer from two major limitations: (1) collecting reference datasets causes expensive annotation cost; (2) the debiasing performance is heavily constrained by the quality of the reference dataset or the additional classifier. To address the above limitations, we propose FairGen, a plug-and-play method that learns attribute latent directions in a self-discovering manner, thus eliminating the reliance on such reference dataset. Specifically, FairGen consists of two parts: a set of attribute adapters and a distribution indicator. Each adapter in the set aims to learn an attribute latent direction, and is optimized via noise composition through a self-discovering process. Then, the distribution indicator is multiplied by the set of adapters to guide the generation process towards the prescribed distribution. Our method enables debiasing multiple attributes in DMs simultaneously, while remaining lightweight and easily integrable with other DMs, eliminating the need for retraining. Extensive experiments on debiasing gender, racial, and their intersectional biases show that our method outperforms previous SOTA by a large margin.
Authors: Jialun Cai, Mengyuan Liu, Hong Liu, Shuheng Zhou, Wenhao Li
Abstract: The widespread application of 3D human pose estimation (HPE) is limited by resource-constrained edge devices, requiring more efficient models. A key approach to enhancing efficiency involves designing networks based on the structural characteristics of input data. However, effectively utilizing the structural priors in human skeletal inputs remains challenging. To address this, we leverage both explicit and implicit spatio-temporal priors of the human body through innovative model design and a pre-training proxy task. First, we propose a Nano Human Topology Network (NanoHTNet), a tiny 3D HPE network with stacked Hierarchical Mixers to capture explicit features. Specifically, the spatial Hierarchical Mixer efficiently learns the human physical topology across multiple semantic levels, while the temporal Hierarchical Mixer with discrete cosine transform and low-pass filtering captures local instantaneous movements and global action coherence. Moreover, Efficient Temporal-Spatial Tokenization (ETST) is introduced to enhance spatio-temporal interaction and reduce computational complexity significantly. Second, PoseCLR is proposed as a general pre-training method based on contrastive learning for 3D HPE, aimed at extracting implicit representations of human topology. By aligning 2D poses from diverse viewpoints in the proxy task, PoseCLR aids 3D HPE encoders like NanoHTNet in more effectively capturing the high-dimensional features of the human body, leading to further performance improvements. Extensive experiments verify that NanoHTNet with PoseCLR outperforms other state-of-the-art methods in efficiency, making it ideal for deployment on edge devices like the Jetson Nano. Code and models are available at https://github.com/vefalun/NanoHTNet.
Authors: Danah Yatim, Rafail Fridman, Omer Bar-Tal, Tali Dekel
Abstract: We present a method for augmenting real-world videos with newly generated dynamic content. Given an input video and a simple user-provided text instruction describing the desired content, our method synthesizes dynamic objects or complex scene effects that naturally interact with the existing scene over time. The position, appearance, and motion of the new content are seamlessly integrated into the original footage while accounting for camera motion, occlusions, and interactions with other dynamic objects in the scene, resulting in a cohesive and realistic output video. We achieve this via a zero-shot, training-free framework that harnesses a pre-trained text-to-video diffusion transformer to synthesize the new content and a pre-trained vision-language model to envision the augmented scene in detail. Specifically, we introduce a novel inference-based method that manipulates features within the attention mechanism, enabling accurate localization and seamless integration of the new content while preserving the integrity of the original scene. Our method is fully automated, requiring only a simple user instruction. We demonstrate its effectiveness on a wide range of edits applied to real-world videos, encompassing diverse objects and scenarios involving both camera and object motion.
Authors: Sharana Dharshikgan Suresh Dass, Hrishav Bakul Barua, Ganesh Krishnasamy, Raveendran Paramesran, Raphael C. -W. Phan
Abstract: Action recognition in dark or low-light (under-exposed) videos is a challenging task due to visibility degradation, which can hinder critical spatiotemporal details. This paper proposes ActLumos, a teacher-student framework that attains single-stream inference while retaining multi-stream level accuracy. The teacher consumes dual stream inputs, which include original dark frames and retinex-enhanced frames, processed by weight-shared R(2+1)D-34 backbones and fused by a Dynamic Feature Fusion (DFF) module, which dynamically re-weights the two streams at each time step, emphasising the most informative temporal segments. The teacher is also included with a supervised contrastive loss (SupCon) that sharpens class margins. The student shares the R(2+1)D-34 backbone but uses only dark frames and no fusion at test time. The student is first pre-trained with self-supervision on dark clips of both datasets without their labels and then fine-tuned with knowledge distillation from the teacher, transferring the teacher's multi-stream knowledge into a single-stream model. Under single-stream inference, the distilled student attains state-of-the-art accuracy of 96.92% (Top-1) on ARID V1.0, 88.27% on ARID V1.5, and 48.96% on Dark48. Ablation studies further highlight the individual contributions of each component, i.e., DFF in the teacher outperforms single or static fusion, knowledge distillation (KD) transfers these gains to the single-stream student, and two-view spatio-temporal SSL surpasses spatial-only or temporal-only variants without increasing inference cost. The official website of this work is available at: https://github.com/HrishavBakulBarua/ActLumos
Authors: Amir Saeidi, Yiran Luo, Agneet Chatterjee, Shamanthak Hegde, Bimsara Pathiraja, Yezhou Yang, Chitta Baral
Abstract: Recent advancements in human preference optimization, originally developed for Large Language Models (LLMs), have shown significant potential in improving text-to-image diffusion models. These methods aim to learn the distribution of preferred samples while distinguishing them from less preferred ones. However, within the existing preference datasets, the original caption often does not clearly favor the preferred image over the alternative, which weakens the supervision signal available during training. To address this issue, we introduce Dual Caption Preference Optimization (DCPO), a data augmentation and optimization framework that reinforces the learning signal by assigning two distinct captions to each preference pair. This encourages the model to better differentiate between preferred and less-preferred outcomes during training. We also construct Pick-Double Caption, a modified version of Pick-a-Pic v2 with separate captions for each image, and propose three different strategies for generating distinct captions: captioning, perturbation, and hybrid methods. Our experiments show that DCPO significantly improves image quality and relevance to prompts, outperforming Stable Diffusion (SD) 2.1, SFT_Chosen, Diffusion-DPO, and MaPO across multiple metrics, including Pickscore, HPSv2.1, GenEval, CLIPscore, and ImageReward, fine-tuned on SD 2.1 as the backbone.
Authors: Guanzhou Ji, Sriram Narayanan, Azadeh Sawyer, Srinivasa Narasimhan
Abstract: This paper introduces a novel image-based rendering technique for jointly estimating indoor lighting and thermal conditions from paired indoor-outdoor high dynamic range (HDR) panoramas. Our method uses the indoor panorama to estimate the 3D floor layout, while the corresponding outdoor panorama serves as an environment map to infer spatially-varying illumination and material properties. Assuming indoor surfaces are Lambertian and that all heat originates from outdoor visible light, we model the relationship between light transport and heat transfer, and perform transient heat simulations to generate indoor temperature distributions. The simulated heat maps are validated against real-world thermal images captured with an infrared camera. This approach supports photorealistic and physically informed visualization, enabling integrated light and heat estimation to advance traditional virtual home staging.
Authors: Nicolas Talabot, Olivier Clerc, Arda Cinar Demirtas, Alexis Goujon, Hieu Le, Doruk Oner, Pascal Fua
Abstract: Accurate 3D shape representation is essential in engineering applications such as design, optimization, and simulation. In practice, engineering workflows require structured, part-based representations, as objects are inherently designed as assemblies of distinct components. However, most existing methods either model shapes holistically or decompose them without predefined part structures, limiting their applicability in real-world design tasks. We propose PartSDF, a supervised implicit representation framework that explicitly models composite shapes with independent, controllable parts while maintaining shape consistency. Thanks to its simple but innovative architecture, PartSDF outperforms both supervised and unsupervised baselines in reconstruction and generation tasks. We further demonstrate its effectiveness as a structured shape prior for engineering applications, enabling precise control over individual components while preserving overall coherence. Code available at https://github.com/cvlab-epfl/PartSDF.
Authors: Guanqi Zhan, Yuanpei Liu, Kai Han, Weidi Xie, Andrew Zisserman
Abstract: The objective in this paper is to improve the performance of text-to-image retrieval. To this end, we introduce a new framework that can boost the performance of large-scale pre-trained vision-language models, so that they can be used for text-to-image re-ranking. The approach, Enhanced Language-Image Pre-training (ELIP), uses the text query, via a simple MLP mapping network, to predict a set of visual prompts to condition the ViT image encoding. ELIP can easily be applied to the commonly used CLIP, SigLIP and BLIP-2 networks. To train the architecture with limited computing resources, we develop a 'student friendly' best practice, involving global hard sample mining, and curation of a large-scale dataset. On the evaluation side, we set up two new out-of-distribution (OOD) benchmarks, Occluded COCO and ImageNet-R, to assess the zero-shot generalisation of the models to different domains. The results demonstrate that ELIP significantly boosts CLIP/SigLIP/SigLIP-2 text-to-image retrieval performance and outperforms BLIP-2 on several benchmarks, as well as providing an easy means to adapt to OOD datasets.
Authors: QingYuan Jiang, Longfei Huang, Yang Yang
Abstract: Multimodal learning (MML) is significantly constrained by modality imbalance, leading to suboptimal performance in practice. While existing approaches primarily focus on balancing the learning of different modalities to address this issue, they fundamentally overlook the inherent disproportion in model classification ability, which serves as the primary cause of this phenomenon. In this paper, we propose a novel multimodal learning approach to dynamically balance the classification ability of weak and strong modalities by incorporating the principle of boosting. Concretely, we first propose a sustained boosting algorithm in multimodal learning by simultaneously optimizing the classification and residual errors. Subsequently, we introduce an adaptive classifier assignment strategy to dynamically facilitate the classification performance of the weak modality. Furthermore, we theoretically analyze the convergence property of the cross-modal gap function, ensuring the effectiveness of the proposed boosting scheme. To this end, the classification ability of strong and weak modalities is expected to be balanced, thereby mitigating the imbalance issue. Empirical experiments on widely used datasets reveal the superiority of our method through comparison with various state-of-the-art (SOTA) multimodal learning baselines. The source code is available at https://github.com/njustkmg/NeurIPS25-AUG.
Authors: Jiageng Zhong, Ming Li, Armin Gruen, Konrad Schindler, Xuan Liao, Qinghua Guo
Abstract: Corals serve as the foundational habitat-building organisms within reef ecosystems, constructing extensive structures that extend over vast distances. However, their inherent fragility and vulnerability to various threats render them susceptible to significant damage and destruction. The application of advanced 3D reconstruction technologies for high-quality modeling is crucial for preserving them. These technologies help scientists to accurately document and monitor the state of coral reefs, including their structure, species distribution and changes over time. Photogrammetry-based approaches stand out among existing solutions, especially with recent advancements in underwater videography, photogrammetric computer vision, and machine learning. Despite continuous progress in image-based 3D reconstruction techniques, there remains a lack of systematic reviews and comprehensive evaluations of cutting-edge solutions specifically applied to underwater coral reef images. The emerging advanced methods may have difficulty coping with underwater imaging environments, complex coral structures, and computational resource constraints. They need to be reviewed and evaluated to bridge the gap between many cutting-edge technical studies and practical applications. This paper focuses on the two critical stages of these approaches: camera pose estimation and dense surface reconstruction. We systematically review and summarize classical and emerging methods, conducting comprehensive evaluations through real-world and simulated datasets. Based on our findings, we offer reference recommendations and discuss the development potential and challenges of existing approaches in depth. This work equips scientists and managers with a technical foundation and practical guidance for processing underwater coral reef images for 3D reconstruction....
Authors: Tianyi Wang, Jianan Fan, Dingxin Zhang, Dongnan Liu, Yong Xia, Heng Huang, Weidong Cai
Abstract: Histopathology and transcriptomics are fundamental modalities in oncology, encapsulating the morphological and molecular aspects of the disease. Multi-modal self-supervised learning has demonstrated remarkable potential in learning pathological representations by integrating diverse data sources. Conventional multi-modal integration methods primarily emphasize modality alignment, while paying insufficient attention to retaining the modality-specific structures. However, unlike conventional scenarios where multi-modal inputs share highly overlapping features, histopathology and transcriptomics exhibit pronounced heterogeneity, offering orthogonal yet complementary insights. Histopathology provides morphological and spatial context, elucidating tissue architecture and cellular topology, whereas transcriptomics delineates molecular signatures through gene expression patterns. This inherent disparity introduces a major challenge in aligning them while maintaining modality-specific fidelity. To address these challenges, we present MIRROR, a novel multi-modal representation learning method designed to foster both modality alignment and retention. MIRROR employs dedicated encoders to extract comprehensive features for each modality, which is further complemented by a modality alignment module to achieve seamless integration between phenotype patterns and molecular profiles. Furthermore, a modality retention module safeguards unique attributes from each modality, while a style clustering module mitigates redundancy and enhances disease-relevant information by modeling and aligning consistent pathological signatures within a clustering space. Extensive evaluations on TCGA cohorts for cancer subtyping and survival analysis highlight MIRROR's superior performance, demonstrating its effectiveness in constructing comprehensive oncological feature representations and benefiting the cancer diagnosis.
Authors: Zhihao Huang, Xi Qiu, Yukuo Ma, Yifu Zhou, Junjie Chen, Hongyuan Zhang, Chi Zhang, Xuelong Li
Abstract: Autoregressive models have achieved significant success in image generation. However, unlike the inherent hierarchical structure of image information in the spectral domain, standard autoregressive methods typically generate pixels sequentially in a fixed spatial order. To better leverage this spectral hierarchy, we introduce NextFrequency Image Generation (NFIG). NFIG is a novel framework that decomposes the image generation process into multiple frequency-guided stages. NFIG aligns the generation process with the natural image structure. It does this by first generating low-frequency components, which efficiently capture global structure with significantly fewer tokens, and then progressively adding higher-frequency details. This frequency-aware paradigm offers substantial advantages: it not only improves the quality of generated images but crucially reduces inference cost by efficiently establishing global structure early on. Extensive experiments on the ImageNet-256 benchmark validate NFIG's effectiveness, demonstrating superior performance (FID: 2.81) and a notable 1.25x speedup compared to the strong baseline VAR-d20. The source code is available at https://github.com/Pride-Huang/NFIG.
Authors: Chiara Cappellino, Gianluca Mancusi, Matteo Mosconi, Angelo Porrello, Simone Calderara, Rita Cucchiara
Abstract: Open-Vocabulary object detectors can generalize to an unrestricted set of categories through simple textual prompting. However, adapting these models to rare classes or reinforcing their abilities on multiple specialized domains remains essential. While recent methods rely on monolithic adaptation strategies with a single set of weights, we embrace modular deep learning. We introduce DitHub, a framework designed to build and maintain a library of efficient adaptation modules. Inspired by Version Control Systems, DitHub manages expert modules as branches that can be fetched and merged as needed. This modular approach allows us to conduct an in-depth exploration of the compositional properties of adaptation modules, marking the first such study in Object Detection. Our method achieves state-of-the-art performance on the ODinW-13 benchmark and ODinW-O, a newly introduced benchmark designed to assess class reappearance. For more details, visit our project page: https://aimagelab.github.io/DitHub/
Authors: Chengxuan Qian, Kai Han, Jianxia Ding, Chongwen Lyu, Zhenlong Yuan, Jun Chen, Zhe Liu
Abstract: Deep learning has shown remarkable success in medical image analysis, but its reliance on large volumes of high-quality labeled data limits its applicability. While noisy labeled data are easier to obtain, directly incorporating them into training can degrade model performance. To address this challenge, we propose a Mean Teacher-based Adaptive Label Correction (ALC) self-ensemble framework for robust medical image segmentation with noisy labels. The framework leverages the Mean Teacher architecture to ensure consistent learning under noise perturbations. It includes an adaptive label refinement mechanism that dynamically captures and weights differences across multiple disturbance versions to enhance the quality of noisy labels. Additionally, a sample-level uncertainty-based label selection algorithm is introduced to prioritize high-confidence samples for network updates, mitigating the impact of noisy annotations. Consistency learning is integrated to align the predictions of the student and teacher networks, further enhancing model robustness. Extensive experiments on two public datasets demonstrate the effectiveness of the proposed framework, showing significant improvements in segmentation performance. By fully exploiting the strengths of the Mean Teacher structure, the ALC framework effectively processes noisy labels, adapts to challenging scenarios, and achieves competitive results compared to state-of-the-art methods.
Authors: Jiyuan Wang, Chunyu Lin, Cheng Guan, Lang Nie, Jing He, Haodong Li, Kang Liao, Yao Zhao
Abstract: In this paper, we propose Jasmine, the first Stable Diffusion (SD)-based self-supervised framework for monocular depth estimation, which effectively harnesses SD's visual priors to enhance the sharpness and generalization of unsupervised prediction. Previous SD-based methods are all supervised since adapting diffusion models for dense prediction requires high-precision supervision. In contrast, self-supervised reprojection suffers from inherent challenges (e.g., occlusions, texture-less regions, illumination variance), and the predictions exhibit blurs and artifacts that severely compromise SD's latent priors. To resolve this, we construct a novel surrogate task of hybrid image reconstruction. Without any additional supervision, it preserves the detail priors of SD models by reconstructing the images themselves while preventing depth estimation from degradation. Furthermore, to address the inherent misalignment between SD's scale and shift invariant estimation and self-supervised scale-invariant depth estimation, we build the Scale-Shift GRU. It not only bridges this distribution gap but also isolates the fine-grained texture of SD output against the interference of reprojection loss. Extensive experiments demonstrate that Jasmine achieves SoTA performance on the KITTI benchmark and exhibits superior zero-shot generalization across multiple datasets.
Authors: Bastian P\"atzold, Jan Nogga, Sven Behnke
Abstract: Vision-language models (VLMs) excel in visual understanding but often lack reliable grounding capabilities and actionable inference rates. Integrating them with open-vocabulary object detection (OVD), instance segmentation, and tracking leverages their strengths while mitigating these drawbacks. We utilize VLM-generated structured descriptions to identify visible object instances, collect application-relevant attributes, and inform an open-vocabulary detector to extract corresponding bounding boxes that are passed to a video segmentation model providing segmentation masks and tracking. Once initialized, this model directly extracts segmentation masks, processing image streams in real time with minimal computational overhead. Tracks can be updated online as needed by generating new structured descriptions and detections. This combines the descriptive power of VLMs with the grounding capability of OVD and the pixel-level understanding and speed of video segmentation. Our evaluation across datasets and robotics platforms demonstrates the broad applicability of this approach, showcasing its ability to extract task-specific attributes from non-standard objects in dynamic environments. Code, data, videos, and benchmarks are available at https://vlm-gist.github.io
Authors: Ziming Wei, Bingqian Lin, Yunshuang Nie, Jiaqi Chen, Shikui Ma, Hang Xu, Xiaodan Liang
Abstract: Data scarcity is a long-standing challenge in the Vision-Language Navigation (VLN) field, which extremely hinders the generalization of agents to unseen environments. Previous works primarily rely on additional simulator data or web-collected images/videos to improve the generalization. However, the simulator environments still face limited diversity, and the web-collected data often requires extensive labor to remove the noise. In this paper, we propose a Rewriting-driven AugMentation (RAM) paradigm for VLN, which directly creates the unseen observation-instruction pairs via rewriting human-annotated training data. Benefiting from our rewriting mechanism, new observation-instruction pairs can be obtained in both simulator-free and labor-saving manners to promote generalization. Specifically, we first introduce Object-Enriched Observation Rewriting, where we combine Vision-Language Models (VLMs) and Large Language Models (LLMs) to derive rewritten object-enriched scene descriptions, enabling observation synthesis with diverse objects and spatial layouts via Text-to-Image Generation Models (T2IMs). Then, we propose Observation-Contrast Instruction Rewriting, which generates observation-aligned rewritten instructions by requiring LLMs to reason the difference between original and new observations. We further develop a mixing-then-focusing training strategy with a random observation cropping scheme, effectively enhancing data distribution diversity while suppressing augmentation data noise during training. Experiments on both the discrete environments (R2R, REVERIE, and R4R datasets) and continuous environments (R2R-CE dataset) show the superior performance and impressive generalization ability of our method. Code is available at https://github.com/SaDil13/VLN-RAM.
Authors: Chenyu Zhang, Daniil Cherniavskii, Antonios Tragoudaras, Antonios Vozikis, Thijmen Nijdam, Derck W. E. Prinzhorn, Mark Bodracska, Nicu Sebe, Andrii Zadaianchuk, Efstratios Gavves
Abstract: Recent advances in image and video generation raise hopes that these models possess world modeling capabilities, the ability to generate realistic, physically plausible videos. This could revolutionize applications in robotics, autonomous driving, and scientific simulation. However, before treating these models as world models, we must ask: Do they adhere to physical conservation laws? To answer this, we introduce Morpheus, a benchmark for evaluating video generation models on physical reasoning. It features 80 real-world videos capturing physical phenomena, guided by conservation laws. Since artificial generations lack ground truth, we assess physical plausibility using physics-informed metrics evaluated with respect to infallible conservation laws known per physical setting, leveraging advances in physics-informed neural networks and vision-language foundation models. Our findings reveal that even with advanced prompting and video conditioning, current models struggle to encode physical principles despite generating aesthetically pleasing videos. All data, leaderboard, and code are open-sourced at our project page.
Authors: Christophe Bolduc, Yannick Hold-Geoffroy, Zhixin Shu, Jean-Fran\c{c}ois Lalonde
Abstract: We present GaSLight, a method that generates spatially-varying lighting from regular images. Our method proposes using HDR Gaussian Splats as light source representation, marking the first time regular images can serve as light sources in a 3D renderer. Our two-stage process first enhances the dynamic range of images plausibly and accurately by leveraging the priors embedded in diffusion models. Next, we employ Gaussian Splats to model 3D lighting, achieving spatially variant lighting. Our approach yields state-of-the-art results on HDR estimations and their applications in illuminating virtual objects and scenes. To facilitate the benchmarking of images as light sources, we introduce a novel dataset of calibrated and unsaturated HDR to evaluate images as light sources. We assess our method using a combination of this novel dataset and an existing dataset from the literature. Project page: https://lvsn.github.io/gaslight/
Authors: Guoqing Zhang, Jingyun Yang, Yang Li
Abstract: Deep learning-based point cloud modeling has been widely investigated as an indispensable component of general shape analysis. Recently, transformer and state space model (SSM) have shown promising capacities in point cloud learning. However, limited research has been conducted on medical point clouds, which have great potential in disease diagnosis and treatment. This paper presents an SSM-based hierarchical feature learning framework for medical point cloud understanding. Specifically, we down-sample input into multiple levels through the farthest point sampling. At each level, we perform a series of k-nearest neighbor (KNN) queries to aggregate multi-scale structural information. To assist SSM in processing point clouds, we introduce coordinate-order and inside-out scanning strategies for efficient serialization of irregular points. Point features are calculated progressively from short neighbor sequences and long point sequences through vanilla and group Point SSM blocks, to capture both local patterns and long-range dependencies. To evaluate the proposed method, we build a large-scale medical point cloud dataset named MedPointS for anatomy classification, completion, and segmentation. Extensive experiments conducted on MedPointS demonstrate that our method achieves superior performance across all tasks. The dataset is available at https://flemme-docs.readthedocs.io/en/latest/medpoints.html. Code is merged to a public medical imaging platform: https://github.com/wlsdzyzl/flemme.
URLs: https://flemme-docs.readthedocs.io/en/latest/medpoints.html., https://github.com/wlsdzyzl/flemme.
Authors: Tsung-Han Wu, Heekyung Lee, Jiaxin Ge, Joseph E. Gonzalez, Trevor Darrell, David M. Chan
Abstract: Vision-Language Models (VLMs) excel at visual understanding but often suffer from visual hallucinations, where they generate descriptions of nonexistent objects, actions, or concepts, posing significant risks in safety-critical applications. Existing hallucination mitigation methods typically follow one of two paradigms: generation adjustment, which modifies decoding behavior to align text with visual inputs, and post-hoc verification, where external models assess and correct outputs. While effective, generation adjustment methods often rely on heuristics and lack correction mechanisms, while post-hoc verification is complicated, typically requiring multiple models and tending to reject outputs rather than refine them. In this work, we introduce REVERSE, a unified framework that integrates hallucination-aware training with on-the-fly self-verification. By leveraging a new hallucination-verification dataset containing over 1.3M semi-synthetic samples, along with a novel inference-time retrospective resampling technique, our approach enables VLMs to both detect hallucinations during generation and dynamically revise those hallucinations. Our evaluations show that REVERSE achieves state-of-the-art hallucination reduction, outperforming the best existing methods by up to 12% on CHAIR-MSCOCO and 34% on HaloQuest. Our dataset, model, and code are available at: https://reverse-vlm.github.io.
Authors: Elena Plekhanova, Damien Robert, Johannes Dollinger, Emilia Arens, Philipp Brun, Jan Dirk Wegner, Niklaus Zimmermann
Abstract: With the exacerbation of the biodiversity and climate crises, macroecological pursuits such as global biodiversity mapping become more urgent. Remote sensing offers a wealth of Earth observation data for ecological studies, but the scarcity of labeled datasets remains a major challenge. Recently, self-supervised learning has enabled learning representations from unlabeled data, triggering the development of pretrained geospatial models with generalizable features. However, these models are often trained on datasets biased toward areas of high human activity, leaving entire ecological regions underrepresented. Additionally, while some datasets attempt to address seasonality through multi-date imagery, they typically follow calendar seasons rather than local phenological cycles. To better capture vegetation seasonality at a global scale, we propose a simple phenology-informed sampling strategy and introduce corresponding SSL4Eco, a multi-date Sentinel-2 dataset, on which we train an existing model with a season-contrastive objective. We compare representations learned from SSL4Eco against other datasets on diverse ecological downstream tasks and demonstrate that our straightforward sampling method consistently improves representation quality, highlighting the importance of dataset construction. The model pretrained on SSL4Eco reaches state of the art performance on 7 out of 8 downstream tasks spanning (multi-label) classification and regression. We release our code, data, and model weights to support macroecological and computer vision research at https://github.com/PlekhanovaElena/ssl4eco.
Authors: Sumit Mamtani, Yash Thesia
Abstract: Fine-grained visual classification aims to recognize objects belonging to many subordinate categories of a supercategory, where appearance alone often fails to distinguish highly similar classes. We propose a unified framework that integrates image, text, and metadata via cross-contrastive pre-training. We first align the three modality encoders in a shared embedding space and then fine-tune the image and metadata encoders for classification. On NABirds, our approach improves over the baseline by 7.83% and achieves 84.44% top-1 accuracy, outperforming strong multimodal methods.
Authors: Deliang Wei, Peng Chen, Haobo Xu, Jiale Yao, Fang Li, Tieyong Zeng
Abstract: Plug-and-play (PnP) methods with deep denoisers have shown impressive results in imaging problems. They typically require strong convexity or smoothness of the fidelity term and a (residual) non-expansive denoiser for convergence. These assumptions, however, are violated in Poisson inverse problems, and non-expansiveness can hinder denoising performance. To address these challenges, we propose a cocoercive conservative (CoCo) denoiser, which may be (residual) expansive, leading to improved denoising. By leveraging the generalized Helmholtz decomposition, we introduce a novel training strategy that combines Hamiltonian regularization to promote conservativeness and spectral regularization to ensure cocoerciveness. We prove that CoCo denoiser is a proximal operator of a weakly convex function, enabling a restoration model with an implicit weakly convex prior. The global convergence of PnP methods to a stationary point of this restoration model is established. Extensive experimental results demonstrate that our approach outperforms closely related methods in both visual quality and quantitative metrics.
Authors: Xuannan Liu, Zekun Li, Zheqi He, Peipei Li, Shuhan Xia, Xing Cui, Huaibo Huang, Xi Yang, Ran He
Abstract: The increasing deployment of Large Vision-Language Models (LVLMs) raises safety concerns under potential malicious inputs. However, existing multimodal safety evaluations primarily focus on model vulnerabilities exposed by static image inputs, ignoring the temporal dynamics of video that may induce distinct safety risks. To bridge this gap, we introduce Video-SafetyBench, the first comprehensive benchmark designed to evaluate the safety of LVLMs under video-text attacks. It comprises 2,264 video-text pairs spanning 48 fine-grained unsafe categories, each pairing a synthesized video with either a harmful query, which contains explicit malice, or a benign query, which appears harmless but triggers harmful behavior when interpreted alongside the video. To generate semantically accurate videos for safety evaluation, we design a controllable pipeline that decomposes video semantics into subject images (what is shown) and motion text (how it moves), which jointly guide the synthesis of query-relevant videos. To effectively evaluate uncertain or borderline harmful outputs, we propose RJScore, a novel LLM-based metric that incorporates the confidence of judge models and human-aligned decision threshold calibration. Extensive experiments show that benign-query video composition achieves average attack success rates of 67.2%, revealing consistent vulnerabilities to video-induced attacks. We believe Video-SafetyBench will catalyze future research into video-based safety evaluation and defense strategies.
Authors: Ziqiang Li, Jiazhen Yan, Ziwen He, Kai Zeng, Weiwei Jiang, Lizhi Xiong, Zhangjie Fu
Abstract: The rapid advancement of generative models, such as GANs and Diffusion models, has enabled the creation of highly realistic synthetic images, raising serious concerns about misinformation, deepfakes, and copyright infringement. Although numerous Artificial Intelligence Generated Image (AIGI) detectors have been proposed, often reporting high accuracy, their effectiveness in real-world scenarios remains questionable. To bridge this gap, we introduce AIGIBench, a comprehensive benchmark designed to rigorously evaluate the robustness and generalization capabilities of state-of-the-art AIGI detectors. AIGIBench simulates real-world challenges through four core tasks: multi-source generalization, robustness to image degradation, sensitivity to data augmentation, and impact of test-time pre-processing. It includes 23 diverse fake image subsets that span both advanced and widely adopted image generation techniques, along with real-world samples collected from social media and AI art platforms. Extensive experiments on 11 advanced detectors demonstrate that, despite their high reported accuracy in controlled settings, these detectors suffer significant performance drops on real-world data, limited benefits from common augmentations, and nuanced effects of pre-processing, highlighting the need for more robust detection strategies. By providing a unified and realistic evaluation framework, AIGIBench offers valuable insights to guide future research toward dependable and generalizable AIGI detection.Data and code are publicly available at: https://github.com/HorizonTEL/AIGIBench.
Authors: Ruichuan An, Sihan Yang, Renrui Zhang, Zijun Shen, Ming Lu, Gaole Dai, Hao Liang, Ziyu Guo, Shilin Yan, Yulin Luo, Bocheng Zou, Chaoqun Yang, Wentao Zhang
Abstract: Personalized models have demonstrated remarkable success in understanding and generating concepts provided by users. However, existing methods use separate concept tokens for understanding and generation, treating these tasks in isolation. This may result in limitations for generating images with complex prompts. For example, given the concept $\langle bo\rangle$, generating "$\langle bo\rangle$ wearing its hat" without additional textual descriptions of its hat. We call this kind of generation \textit{\textbf{personalized attribute-reasoning generation}}. To address the limitation, we present UniCTokens, a novel framework that effectively integrates personalized information into a unified vision language model (VLM) for understanding and generation. UniCTokens trains a set of unified concept tokens to leverage complementary semantics, boosting two personalized tasks. Moreover, we propose a progressive training strategy with three stages: understanding warm-up, bootstrapping generation from understanding, and deepening understanding from generation to enhance mutual benefits between both tasks. To quantitatively evaluate the unified VLM personalization, we present UnifyBench, the first benchmark for assessing concept understanding, concept generation, and attribute-reasoning generation. Experimental results on UnifyBench indicate that UniCTokens shows competitive performance compared to leading methods in concept understanding, concept generation, and achieving state-of-the-art results in personalized attribute-reasoning generation. Our research demonstrates that enhanced understanding improves generation, and the generation process can yield valuable insights into understanding. Our code and dataset will be released at: \href{https://github.com/arctanxarc/UniCTokens}{https://github.com/arctanxarc/UniCTokens}.
URLs: https://github.com/arctanxarc/UniCTokens, https://github.com/arctanxarc/UniCTokens
Authors: Ming Yang, Haoran Li
Abstract: 6DoF object pose estimation is fundamental to robotic grasp tasks. While recent learning-based methods achieve high accuracy, their computational demands hinder deployment on resource-constrained mobile platforms. In this work, we revisit the classical keypoint matching paradigm and propose GMatch, a lightweight, geometry-constrained keypoint matcher that can run efficiently on embedded CPU-only platforms. GMatch works with keypoint descriptors and it uses a set of geometric constraints to establishes inherent ambiguities between features extracted by descriptors, thus giving a globally consistent correspondences from which 6DoF pose can be easily solved. We benchmark GMatch on the HOPE and YCB-Video datasets, where our method beats existing keypoint matchers (both feature-based and geometry-based) among three commonly used descriptors and approaches the SOTA zero-shot method on texture-rich objects with much more humble devices. The method is further deployed on a LoCoBot mobile manipulator, enabling a one-shot grasp pipeline that demonstrates high task success rates in real-world experiments. In a word, by its lightweight and white-box nature, GMatch offers a practical solution for resource-limited robotic systems, and although currently bottlenecked by descriptor quality, the framework presents a promising direction towards robust yet efficient pose estimation. Code will be released soon under Mozilla Public License.
Authors: Wenjin Qin, Hailin Wang, Hao Shu, Feng Zhang, Jianjun Wang, Xiangyong Cao, Xi-Le Zhao, Gemine Vivone
Abstract: In recent years, tensor decomposition-based approaches for hyperspectral anomaly detection (HAD) have gained significant attention in the field of remote sensing. However, existing methods often fail to fully leverage both the global correlations and local smoothness of the background components in hyperspectral images (HSIs), which exist in both the spectral and spatial domains. This limitation results in suboptimal detection performance. To mitigate this critical issue, we put forward a novel HAD method named HAD-EUNTRFR, which incorporates an enhanced unified nonconvex tensor ring (TR) factors regularization. In the HAD-EUNTRFR framework, the raw HSIs are first decomposed into background and anomaly components. The TR decomposition is then employed to capture the spatial-spectral correlations within the background component. Additionally, we introduce a unified and efficient nonconvex regularizer, induced by tensor singular value decomposition (TSVD), to simultaneously encode the low-rankness and sparsity of the 3-D gradient TR factors into a unique concise form. The above characterization scheme enables the interpretable gradient TR factors to inherit the low-rankness and smoothness of the original background. To further enhance anomaly detection, we design a generalized nonconvex regularization term to exploit the group sparsity of the anomaly component. To solve the resulting doubly nonconvex model, we develop a highly efficient optimization algorithm based on the alternating direction method of multipliers (ADMM) framework. Experimental results on several benchmark datasets demonstrate that our proposed method outperforms existing state-of-the-art (SOTA) approaches in terms of detection accuracy.
Authors: Peng Wang, Xiang Liu, Peidong Liu
Abstract: Stylizing 3D scenes instantly while maintaining multi-view consistency and faithfully resembling a style image remains a significant challenge. Current state-of-the-art 3D stylization methods typically involve computationally intensive test-time optimization to transfer artistic features into a pretrained 3D representation, often requiring dense posed input images. In contrast, leveraging recent advances in feed-forward reconstruction models, we demonstrate a novel approach to achieve direct 3D stylization in less than a second using unposed sparse-view scene images and an arbitrary style image. To address the inherent decoupling between reconstruction and stylization, we introduce a branched architecture that separates structure modeling and appearance shading, effectively preventing stylistic transfer from distorting the underlying 3D scene structure. Furthermore, we adapt an identity loss to facilitate pre-training our stylization model through the novel view synthesis task. This strategy also allows our model to retain its original reconstruction capabilities while being fine-tuned for stylization. Comprehensive evaluations, using both in-domain and out-of-domain datasets, demonstrate that our approach produces high-quality stylized 3D content that achieve a superior blend of style and scene appearance, while also outperforming existing methods in terms of multi-view consistency and efficiency.
Authors: Matthew Beveridge, Shree K. Nayar
Abstract: We introduce a taxonomy of materials for hierarchical recognition from local appearance. Our taxonomy is motivated by vision applications and is arranged according to the physical traits of materials. We contribute a diverse, in-the-wild dataset with images and depth maps of the taxonomy classes. Utilizing the taxonomy and dataset, we present a method for hierarchical material recognition based on graph attention networks. Our model leverages the taxonomic proximity between classes and achieves state-of-the-art performance. We demonstrate the model's potential to generalize to adverse, real-world imaging conditions, and that novel views rendered using the depth maps can enhance this capability. Finally, we show the model's capacity to rapidly learn new materials in a few-shot learning setting.
Authors: Gabriel Sarch, Snigdha Saha, Naitik Khandelwal, Ayush Jain, Michael J. Tarr, Aviral Kumar, Katerina Fragkiadaki
Abstract: While reinforcement learning (RL) over chains of thought has significantly advanced language models in tasks such as mathematics and coding, visual reasoning introduces added complexity by requiring models to direct visual attention, interpret perceptual inputs, and ground abstract reasoning in spatial evidence. We introduce ViGoRL (Visually Grounded Reinforcement Learning), a vision-language model trained with RL to explicitly anchor each reasoning step to specific visual coordinates. Inspired by human visual decision-making, ViGoRL learns to produce spatially grounded reasoning traces, guiding visual attention to task-relevant regions at each step. When fine-grained exploration is required, our novel multi-turn RL framework enables the model to dynamically zoom into predicted coordinates as reasoning unfolds. Across a diverse set of visual reasoning benchmarks--including SAT-2 and BLINK for spatial reasoning, V*bench for visual search, and ScreenSpot and VisualWebArena for web-based grounding--ViGoRL consistently outperforms both supervised fine-tuning and conventional RL baselines that lack explicit grounding mechanisms. Incorporating multi-turn RL with zoomed-in visual feedback significantly improves ViGoRL's performance on localizing small GUI elements and visual search, achieving 86.4% on V*Bench. Additionally, we find that grounding amplifies other visual behaviors such as region exploration, grounded subgoal setting, and visual verification. Finally, human evaluations show that the model's visual references are not only spatially accurate but also helpful for understanding model reasoning steps. Our results show that visually grounded RL is a strong paradigm for imbuing models with general-purpose visual reasoning.
Authors: Ke Niu, Zhuofan Chen, Haiyang Yu, Yuwen Chen, Teng Fu, Mengyang Zhao, Bin Li, Xiangyang Xue
Abstract: Computer-Aided Design (CAD) plays a pivotal role in industrial manufacturing. Orthographic projection reasoning underpins the entire CAD workflow, encompassing design, manufacturing, and simulation. However, prevailing deep-learning approaches employ standard 3D reconstruction pipelines as an alternative, which often introduce imprecise dimensions and limit the parametric editability required for CAD workflows. Recently, some researchers adopt vision-language models (VLMs), particularly supervised fine-tuning (SFT), to tackle CAD-related challenges. SFT shows promise but often devolves into pattern memorization, yielding poor out-of-distribution performance on complex reasoning tasks. To address these gaps, we introduce CReFT-CAD, a two-stage fine-tuning paradigm that first employs a curriculum-driven reinforcement learning stage with difficulty-aware rewards to build reasoning ability steadily, and then applies supervised post-tuning to hone instruction following and semantic extraction. Complementing this, we release TriView2CAD, the first large-scale, open-source benchmark for orthographic projection reasoning, comprising 200,000 synthetic and 3,000 real-world orthographic projections with precise dimension annotations and six interoperable data modalities. We benchmark leading VLMs on orthographic projection reasoning and demonstrate that CReFT-CAD substantially improves reasoning accuracy and out-of-distribution generalizability in real-world scenarios, offering valuable insights for advancing CAD reasoning research.
Authors: Hao Yan, Handong Zheng, Hao Wang, Liang Yin, Xingchen Liu, Zhenbiao Cao, Xinxing Su, Zihao Chen, Jihao Wu, Minghui Liao, Chao Weng, Wei Chen, Yuliang Liu, Xiang Bai
Abstract: Recent strides in multimodal large language models (MLLMs) have significantly advanced their performance in many reasoning tasks. However, Abstract Visual Reasoning (AVR) remains a critical challenge, primarily due to limitations in perceiving abstract graphics. To tackle this issue, we investigate the bottlenecks in current MLLMs and synthesize training data to improve their abstract visual perception. First, we propose VisuRiddles, a benchmark for AVR, featuring tasks meticulously constructed to assess models' reasoning capacities across five core dimensions and two high-level reasoning categories. Second, we introduce the Perceptual Riddle Synthesizer (PRS), an automated framework for generating riddles with fine-grained perceptual descriptions. PRS not only generates valuable training data for abstract graphics but also provides fine-grained perceptual description, crucially allowing for supervision over intermediate reasoning stages and thereby improving both training efficacy and model interpretability. Our extensive experimental results on VisuRiddles empirically validate that fine-grained visual perception is the principal bottleneck and our synthesis framework markedly enhances the performance of contemporary MLLMs on these challenging tasks. Our code and dataset will be released at https://github.com/yh-hust/VisuRiddles
Authors: Baode Wang, Biao Wu, Weizhen Li, Meng Fang, Zuming Huang, Jun Huang, Haozhe Wang, Yanjie Liang, Ling Chen, Wei Chu, Yuan Qi
Abstract: Automated parsing of scanned documents into richly structured, machine-readable formats remains a critical bottleneck in Document AI, as traditional multi-stage pipelines suffer from error propagation and limited adaptability to diverse layouts. We introduce layoutRL, an end-to-end reinforcement learning framework that trains models to be explicitly layout-aware by optimizing a composite reward of normalized edit distance, paragraph count accuracy, and reading order preservation. Leveraging our newly released dataset, Infinity-Doc-55K, which combines 55K high-fidelity synthetic scanned document parsing data with expert-filtered real-world documents, we instantiate layoutRL in a vision-language-model-based parser called Infinity-Parser. Evaluated on English and Chinese benchmarks for OCR, table and formula extraction, and reading order detection, Infinity-Parser achieves new state-of-the-art performance in both accuracy and structural fidelity, outpacing specialist pipelines and general-purpose vision-language models. We will publicly release our code and dataset to accelerate progress in robust document understanding.
Authors: Yunhao Gou, Kai Chen, Zhili Liu, Lanqing Hong, Xin Jin, Zhenguo Li, James T. Kwok, Yu Zhang
Abstract: Recent breakthroughs in reasoning language models have significantly advanced text-based reasoning. On the other hand, Multi-modal Large Language Models (MLLMs) still lag behind, hindered by their outdated internal LLMs. Upgrading these is often prohibitively expensive, as it requires complete vision-language alignment retraining which is costly. To address this issue, we introduce Perception-Reasoning Decoupling, which modularizes the MLLM's reasoning component and makes it easily replaceable. This approach redefines the MLLM's role to convert multi-modal inputs into detailed textual outputs that can be processed by any powerful, external, text-only LLM reasoners. To align the MLLM's perceptual output with the final reasoning task, we propose a novel reinforcement learning algorithm called Visual Perception Optimization (VPO). VPO rewards the MLLM based on the correctness of answers generated by the external reasoner to produce faithful and query-relevant captions. Together, this decoupling pipeline and VPO form our Reasoning-Aligned PerceptIon Decoupling (RAPID) approach. Empirical results show that RAPID achieves significant performance gains on multi-modal reasoning benchmarks. Crucially, RAPID enables a novel inference-time scaling paradigm: Once trained with VPO, the MLLM can be paired with any state-of-the-art LLM reasoner for consistent performance improvement without retraining.
Authors: Mingxiao Li, Mang Ning, Marie-Francine Moens
Abstract: Text-to-image generation models have made significant progress in producing high-quality images from textual descriptions, yet they continue to struggle with maintaining subject consistency across multiple images, a fundamental requirement for visual storytelling. Existing methods attempt to address this by either fine-tuning models on large-scale story visualization datasets, which is resource-intensive, or by using training-free techniques that share information across generations, which still yield limited success. In this paper, we introduce a novel training-free sampling strategy called Zigzag Sampling with Asymmetric Prompts and Visual Sharing to enhance subject consistency in visual story generation. Our approach proposes a zigzag sampling mechanism that alternates between asymmetric prompting to retain subject characteristics, while a visual sharing module transfers visual cues across generated images to %further enforce consistency. Experimental results, based on both quantitative metrics and qualitative evaluations, demonstrate that our method significantly outperforms previous approaches in generating coherent and consistent visual stories. The code is available at https://github.com/Mingxiao-Li/Asymmetry-Zigzag-StoryDiffusion.
URLs: https://github.com/Mingxiao-Li/Asymmetry-Zigzag-StoryDiffusion.
Authors: Zhanwei Zhang, Kaiyuan Liu, Junjie Liu, Wenxiao Wang, Binbin Lin, Liang Xie, Chen Shen, Deng Cai
Abstract: Local geometry-controllable computer-aided design (CAD) generation aims to modify local parts of CAD models automatically, enhancing design efficiency. It also ensures that the shapes of newly generated local parts follow user-specific geometric instructions (e.g., an isosceles right triangle or a rectangle with one corner cut off). However, existing methods encounter challenges in achieving this goal. Specifically, they either lack the ability to follow textual instructions or are unable to focus on the local parts. To address this limitation, we introduce GeoCAD, a user-friendly and local geometry-controllable CAD generation method. Specifically, we first propose a complementary captioning strategy to generate geometric instructions for local parts. This strategy involves vertex-based and VLLM-based captioning for systematically annotating simple and complex parts, respectively. In this way, we caption $\sim$221k different local parts in total. In the training stage, given a CAD model, we randomly mask a local part. Then, using its geometric instruction and the remaining parts as input, we prompt large language models (LLMs) to predict the masked part. During inference, users can specify any local part for modification while adhering to a variety of predefined geometric instructions. Extensive experiments demonstrate the effectiveness of GeoCAD in generation quality, validity and text-to-CAD consistency. Code will be available at https://github.com/Zhanwei-Z/GeoCAD.
Authors: Yanlong Chen, Mattia Orlandi, Pierangelo Maria Rapa, Simone Benatti, Luca Benini, Yawei Li
Abstract: Human-machine interaction, particularly in prosthetic and robotic control, has seen progress with gesture recognition via surface electromyographic (sEMG) signals.However, classifying similar gestures that produce nearly identical muscle signals remains a challenge, often reducing classification accuracy. Traditional deep learning models for sEMG gesture recognition are large and computationally expensive, limiting their deployment on resource-constrained embedded systems. In this work, we propose WaveFormer, a lightweight transformer-based architecture tailored for sEMG gesture recognition. Our model integrates time-domain and frequency-domain features through a novel learnable wavelet transform, enhancing feature extraction. In particular, the WaveletConv module, a multi-level wavelet decomposition layer with depthwise separable convolution, ensures both efficiency and compactness. With just 3.1 million parameters, WaveFormer achieves 95% classification accuracy on the EPN612 dataset, outperforming larger models. Furthermore, when profiled on a laptop equipped with an Intel CPU, INT8 quantization achieves real-time deployment with a 6.75 ms inference latency.
Authors: Jingfeng Guo, Jian Liu, Jinnan Chen, Shiwei Mao, Changrong Hu, Puhua Jiang, Junlin Yu, Jing Xu, Qi Liu, Lixin Xu, Zhuo Chen, Chunchao Guo
Abstract: We introduce Auto-Connect, a novel approach for automatic rigging that explicitly preserves skeletal connectivity through a connectivity-preserving tokenization scheme. Unlike previous methods that predict bone positions represented as two joints or first predict points before determining connectivity, our method employs special tokens to define endpoints for each joint's children and for each hierarchical layer, effectively automating connectivity relationships. This approach significantly enhances topological accuracy by integrating connectivity information directly into the prediction framework. To further guarantee high-quality topology, we implement a topology-aware reward function that quantifies topological correctness, which is then utilized in a post-training phase through reward-guided Direct Preference Optimization. Additionally, we incorporate implicit geodesic features for latent top-k bone selection, which substantially improves skinning quality. By leveraging geodesic distance information within the model's latent space, our approach intelligently determines the most influential bones for each vertex, effectively mitigating common skinning artifacts. This combination of connectivity-preserving tokenization, reward-guided fine-tuning, and geodesic-aware bone selection enables our model to consistently generate more anatomically plausible skeletal structures with superior deformation properties.
Authors: Yujun Wang, Aniri, Jinhe Bi, Soeren Pirk, Yunpu Ma
Abstract: Multimodal large language models (MLLMs) frequently hallucinate by over-committing to spurious visual cues. Prior remedies-Visual and Instruction Contrastive Decoding (VCD, ICD)-mitigate this issue, yet the mechanism remains opaque. We first empirically show that their improvements systematically coincide with redistributions of cross-modal attention. Building on this insight, we propose Attention-Steerable Contrastive Decoding (ASCD), which directly steers the attention scores during decoding. ASCD combines (i) positive steering, which amplifies automatically mined text-centric heads-stable within a model and robust across domains-with (ii) negative steering, which dampens on-the-fly identified critical visual tokens. The method incurs negligible runtime and memory overhead and requires no additional training. Across five MLLM backbones and three decoding schemes, ASCD reduces hallucination on POPE, CHAIR, and MMHal-Bench by up to 38.2 percent while improving accuracy on standard VQA benchmarks, including MMMU, MM-VET, ScienceQA, TextVQA, and GQA. These results position attention steering as a simple, model-agnostic, and principled route to safer, more faithful multimodal generation.
Authors: Tao Liu, Dafeng Zhang, Gengchen Li, Shizhuo Liu, Yongqi Song, Senmao Li, Shiqi Yang, Boqian Li, Kai Wang, Yaxing Wang
Abstract: Face aging has become a crucial task in computer vision, with applications ranging from entertainment to healthcare. However, existing methods struggle with achieving a realistic and seamless transformation across the entire lifespan, especially when handling large age gaps or extreme head poses. The core challenge lies in balancing age accuracy and identity preservation--what we refer to as the Age-ID trade-off. Most prior methods either prioritize age transformation at the expense of identity consistency or vice versa. In this work, we address this issue by proposing a two-pass face aging framework, named Cradle2Cane, based on few-step text-to-image (T2I) diffusion models. The first pass focuses on solving age accuracy by introducing an adaptive noise injection (AdaNI) mechanism. This mechanism is guided by including prompt descriptions of age and gender for the given person as the textual condition. Also, by adjusting the noise level, we can control the strength of aging while allowing more flexibility in transforming the face. However, identity preservation is weakly ensured here to facilitate stronger age transformations. In the second pass, we enhance identity preservation while maintaining age-specific features by conditioning the model on two identity-aware embeddings (IDEmb): SVR-ArcFace and Rotate-CLIP. This pass allows for denoising the transformed image from the first pass, ensuring stronger identity preservation without compromising the aging accuracy. Both passes are jointly trained in an end-to-end way. Extensive experiments on the CelebA-HQ test dataset, evaluated through Face++ and Qwen-VL protocols, show that our Cradle2Cane outperforms existing face aging methods in age accuracy and identity consistency. Code is available at https://github.com/byliutao/Cradle2Cane.
Authors: Pufan Li, Bi'an Du, Wei Hu
Abstract: Generating realistic 3D objects from single-view images requires natural appearance, 3D consistency, and the ability to capture multiple plausible interpretations of unseen regions. Existing approaches often rely on fine-tuning pretrained 2D diffusion models or directly generating 3D information through fast network inference or 3D Gaussian Splatting, but their results generally suffer from poor multiview consistency and lack geometric detail. To tackle these issues, we present a novel method that seamlessly integrates geometry and perception information without requiring additional model training to reconstruct detailed 3D objects from a single image. Specifically, we incorporate geometry and perception priors to initialize the Gaussian branches and guide their parameter optimization. The geometry prior captures the rough 3D shapes, while the perception prior utilizes the 2D pretrained diffusion model to enhance multiview information. Subsequently, we introduce a stable Score Distillation Sampling for fine-grained prior distillation to ensure effective knowledge transfer. The model is further enhanced by a reprojection-based strategy that enforces depth consistency. Experimental results show that we outperform existing methods on novel view synthesis and 3D reconstruction, demonstrating robust and consistent 3D object generation.
Authors: Mohammed Rakib, Arunkumar Bagavathi
Abstract: Multimodal learning aims to leverage information from diverse data modalities to achieve more comprehensive performance. However, conventional multimodal models often suffer from modality imbalance, where one or a few modalities dominate model optimization, leading to suboptimal feature representation and underutilization of weak modalities. To address this challenge, we introduce Gradient-Guided Distillation (G$^{2}$D), a knowledge distillation framework that optimizes the multimodal model with a custom-built loss function that fuses both unimodal and multimodal objectives. G$^{2}$D further incorporates a dynamic sequential modality prioritization (SMP) technique in the learning process to ensure each modality leads the learning process, avoiding the pitfall of stronger modalities overshadowing weaker ones. We validate G$^{2}$D on multiple real-world datasets and show that G$^{2}$D amplifies the significance of weak modalities while training and outperforms state-of-the-art methods in classification and regression tasks. Our code is available at https://github.com/rAIson-Lab/G2D.
Authors: Amir Aghdam, Vincent Tao Hu, Bj\"orn Ommer
Abstract: We address the task of zero-shot video classification for extremely fine-grained actions (e.g., Windmill Dunk in basketball), where no video examples or temporal annotations are available for unseen classes. While image-language models (e.g., CLIP, SigLIP) show strong open-set recognition, they lack temporal modeling needed for video understanding. We propose ActAlign, a truly zero-shot, training-free method that formulates video classification as a sequence alignment problem, preserving the generalization strength of pretrained image-language models. For each class, a large language model (LLM) generates an ordered sequence of sub-actions, which we align with video frames using Dynamic Time Warping (DTW) in a shared embedding space. Without any video-text supervision or fine-tuning, ActAlign achieves 30.5% accuracy on ActionAtlas--the most diverse benchmark of fine-grained actions across multiple sports--where human performance is only 61.6%. ActAlign outperforms billion-parameter video-language models while using 8x fewer parameters. Our approach is model-agnostic and domain-general, demonstrating that structured language priors combined with classical alignment methods can unlock the open-set recognition potential of image-language models for fine-grained video understanding.
Authors: Bingfan Zhu, Biao Jiang, Sunyi Wang, Shixiang Tang, Tao Chen, Linjie Luo, Youyi Zheng, Xin Chen
Abstract: With the rapid progress of large language models (LLMs), multimodal frameworks that unify understanding and generation have become promising, yet they face increasing complexity as the number of modalities and tasks grows. We observe that motion quantization introduces approximation errors that cap motion quality, and that unifying discrete text and continuous motion within a single-stream backbone amplifies cross-modal interference. Motivated by recent multi-branch Transformer designs that separate signals from different modalities, we propose MotionGPT3, a bimodal motion-language model for both understanding and generation. MotionGPT3 encodes raw motion into a continuous latent space using a variational autoencoder (VAE), thereby avoiding quantization-induced artifacts, while leveraging the semantic prior of pretrained language models. A dual-stream Transformer with shared attention preserves modality-specific routes while enabling controlled, bidirectional information flow, which reduces interference, stabilizing optimization, and empirically accelerates convergence without degrading fidelity. For multimodal joint training, a generate-then-align three-stage schedule further improves stability and limits cross-task interference. Experiments show that MotionGPT3 achieves 2x faster convergence in training loss and up to 4x faster convergence in validation, while maintaining state-of-the-art performance on standard motion understanding and motion generation benchmarks.
Authors: Christian Intern\`o, Robert Geirhos, Markus Olhofer, Sunny Liu, Barbara Hammer, David Klindt
Abstract: The rapid advancement of generative AI enables highly realistic synthetic videos, posing significant challenges for content authentication and raising urgent concerns about misuse. Existing detection methods often struggle with generalization and capturing subtle temporal inconsistencies. We propose ReStraV(Representation Straightening Video), a novel approach to distinguish natural from AI-generated videos. Inspired by the "perceptual straightening" hypothesis -- which suggests real-world video trajectories become more straight in neural representation domain -- we analyze deviations from this expected geometric property. Using a pre-trained self-supervised vision transformer (DINOv2), we quantify the temporal curvature and stepwise distance in the model's representation domain. We aggregate statistics of these measures for each video and train a classifier. Our analysis shows that AI-generated videos exhibit significantly different curvature and distance patterns compared to real videos. A lightweight classifier achieves state-of-the-art detection performance (e.g., 97.17% accuracy and 98.63% AUROC on the VidProM benchmark), substantially outperforming existing image- and video-based methods. ReStraV is computationally efficient, it is offering a low-cost and effective detection solution. This work provides new insights into using neural representation geometry for AI-generated video detection.
Authors: Lin Wu, Zhixiang Chen, Jianglin Lan
Abstract: Generating realistic 3D human-object interactions (HOIs) remains a challenging task due to the difficulty of modeling detailed interaction dynamics. Existing methods treat human and object motions independently, resulting in physically implausible and causally inconsistent behaviors. In this work, we present HOI-Dyn, a novel framework that formulates HOI generation as a driver-responder system, where human actions drive object responses. At the core of our method is a lightweight transformer-based interaction dynamics model that explicitly predicts how objects should react to human motion. To further enforce consistency, we introduce a residual-based dynamics loss that mitigates the impact of dynamics prediction errors and prevents misleading optimization signals. The dynamics model is used only during training, preserving inference efficiency. Through extensive qualitative and quantitative experiments, we demonstrate that our approach not only enhances the quality of HOI generation but also establishes a feasible metric for evaluating the quality of generated interactions.
Authors: Jingyao Wang, Yiming Chen, Lingyu Si, Changwen Zheng
Abstract: Scene understanding is one of the core tasks in computer vision, aiming to extract semantic information from images to identify objects, scene categories, and their interrelationships. Although advancements in Vision-Language Models (VLMs) have driven progress in this field, existing VLMs still face challenges in adaptation to unseen complex wide-area scenes. To address the challenges, this paper proposes a Hierarchical Coresets Selection (HCS) mechanism to advance the adaptation of VLMs in complex wide-area scene understanding. It progressively refines the selected regions based on the proposed theoretically guaranteed importance function, which considers utility, representativeness, robustness, and synergy. Without requiring additional fine-tuning, HCS enables VLMs to achieve rapid understandings of unseen scenes at any scale using minimal interpretable regions while mitigating insufficient feature density. HCS is a plug-and-play method that is compatible with any VLM. Experiments demonstrate that HCS achieves superior performance and universality in various tasks.
Authors: Peirong Zhang, Haowei Xu, Jiaxin Zhang, Guitao Xu, Xuhan Zheng, Zhenhua Yang, Junle Liu, Yuyi Zhang, Lianwen Jin
Abstract: Text image is a unique and crucial information medium that integrates visual aesthetics and linguistic semantics in modern e-society. Due to their subtlety and complexity, the generation of text images represents a challenging and evolving frontier in the image generation field. The recent surge of specialized image generators (\emph{e.g.}, Flux-series) and unified generative models (\emph{e.g.}, GPT-4o), which demonstrate exceptional fidelity, raises a natural question: can they master the intricacies of text image generation and editing? Motivated by this, we assess current state-of-the-art generative models' capabilities in terms of text image generation and editing. We incorporate various typical optical character recognition (OCR) tasks into our evaluation and broaden the concept of text-based generation tasks into OCR generative tasks. We select 33 representative tasks and categorize them into five categories: document, handwritten text, scene text, artistic text, and complex \& layout-rich text. For comprehensive evaluation, we examine six models across both closed-source and open-source domains, using tailored, high-quality image inputs and prompts. Through this evaluation, we draw crucial observations and identify the weaknesses of current generative models for OCR tasks. We argue that photorealistic text image generation and editing should be internalized as foundational skills into general-domain generative models, rather than being delegated to specialized solutions, and we hope this empirical analysis can provide valuable insights for the community to achieve this goal. This evaluation is online and will be continuously updated at our GitHub repository.
Authors: Yotam Erel, Olaf D\"unkel, Rishabh Dabral, Vladislav Golyanik, Christian Theobalt, Amit H. Bermano
Abstract: We introduce a new interpretation of the attention matrix as a discrete-time Markov chain. Our interpretation sheds light on common operations involving attention scores such as selection, summation, and averaging in a unified framework. It further extends them by considering indirect attention, propagated through the Markov chain, as opposed to previous studies that only model immediate effects. Our key observation is that tokens linked to semantically similar regions form metastable states, i.e., regions where attention tends to concentrate, while noisy attention scores dissipate. Metastable states and their prevalence can be easily computed through simple matrix multiplication and eigenanalysis, respectively. Using these lightweight tools, we demonstrate state-of-the-art zero-shot segmentation. Lastly, we define TokenRank -- the steady state vector of the Markov chain, which measures global token importance. We show that TokenRank enhances unconditional image generation, improving both quality (IS) and diversity (FID), and can also be incorporated into existing segmentation techniques to improve their performance over existing benchmarks. We believe our framework offers a fresh view of how tokens are being attended in modern visual transformers.
Authors: Chengxuan Zhu, Qingnan Fan, Qi Zhang, Jinwei Chen, Huaqi Zhang, Chao Xu, Boxin Shi
Abstract: We introduce BokehDiff, a novel lens blur rendering method that achieves physically accurate and visually appealing outcomes, with the help of generative diffusion prior. Previous methods are bounded by the accuracy of depth estimation, generating artifacts in depth discontinuities. Our method employs a physics-inspired self-attention module that aligns with the image formation process, incorporating depth-dependent circle of confusion constraint and self-occlusion effects. We adapt the diffusion model to the one-step inference scheme without introducing additional noise, and achieve results of high quality and fidelity. To address the lack of scalable paired data, we propose to synthesize photorealistic foregrounds with transparency with diffusion models, balancing authenticity and scene diversity.
Authors: Yilei Jiang, Yaozhi Zheng, Yuxuan Wan, Jiaming Han, Qunzhong Wang, Michael R. Lyu, Xiangyu Yue
Abstract: Automating the transformation of user interface (UI) designs into front-end code holds significant promise for accelerating software development and democratizing design workflows. While multimodal large language models (MLLMs) can translate images to code, they often fail on complex UIs, struggling to unify visual perception, layout planning, and code synthesis within a single monolithic model, which leads to frequent perception and planning errors. To address this, we propose ScreenCoder, a modular multi-agent framework that decomposes the task into three interpretable stages: grounding, planning, and generation. By assigning these distinct responsibilities to specialized agents, our framework achieves significantly higher robustness and fidelity than end-to-end approaches. Furthermore, ScreenCoder serves as a scalable data engine, enabling us to generate high-quality image-code pairs. We use this data to fine-tune open-source MLLM via a dual-stage pipeline of supervised fine-tuning and reinforcement learning, demonstrating substantial gains in its UI generation capabilities. Extensive experiments demonstrate that our approach achieves state-of-the-art performance in layout accuracy, structural coherence, and code correctness. Our code is made publicly available at https://github.com/leigest519/ScreenCoder.
Authors: Alexandre Brown, Glen Berseth
Abstract: Visual reinforcement learning (RL) is challenging due to the need to extract useful representations from high-dimensional inputs while learning effective control from sparse and noisy rewards. Although large perception models exist, integrating them effectively into RL for visual generalization and improved sample efficiency remains difficult. We propose SegDAC, a Segmentation-Driven Actor-Critic method. SegDAC uses Segment Anything (SAM) for object-centric decomposition and YOLO-World to ground the image segmentation process via text inputs. It includes a novel transformer-based architecture that supports a dynamic number of segments at each time step and effectively learns which segments to focus on using online RL, without using human labels. By evaluating SegDAC over a challenging visual generalization benchmark using Maniskill3, which covers diverse manipulation tasks under strong visual perturbations, we demonstrate that SegDAC achieves significantly better visual generalization, doubling prior performance on the hardest setting and matching or surpassing prior methods in sample efficiency across all evaluated tasks.
Authors: Haidong Xu, Guangwei Xu, Zhedong Zheng, Xiatian Zhu, Wei Ji, Xiangtai Li, Ruijie Guo, Meishan Zhang, Min zhang, Hao Fei
Abstract: This paper introduces VimoRAG, a novel video-based retrieval-augmented motion generation framework for motion large language models (LLMs). As motion LLMs face severe out-of-domain/out-of-vocabulary issues due to limited annotated data, VimoRAG leverages large-scale in-the-wild video databases to enhance 3D motion generation by retrieving relevant 2D human motion signals. While video-based motion RAG is nontrivial, we address two key bottlenecks: (1) developing an effective motion-centered video retrieval model that distinguishes human poses and actions, and (2) mitigating the issue of error propagation caused by suboptimal retrieval results. We design the Gemini Motion Video Retriever mechanism and the Motion-centric Dual-alignment DPO Trainer, enabling effective retrieval and generation processes. Experimental results show that VimoRAG significantly boosts the performance of motion LLMs constrained to text-only input. All the resources are available at https://walkermitty.github.io/VimoRAG/
Authors: Mona Mirzaie, Bodo Rosenhahn
Abstract: Trustworthy AI is mandatory for the broad deployment of autonomous vehicles. Although end-to-end approaches derive control commands directly from raw data, interpreting these decisions remains challenging, especially in complex urban scenarios. This is mainly attributed to very deep neural networks with non-linear decision boundaries, making it challenging to grasp the logic behind AI-driven decisions. This paper presents a method to enhance interpretability while optimizing control commands in autonomous driving. To address this, we propose loss functions that promote the interpretability of our model by generating sparse and localized feature maps. The feature activations allow us to explain which image regions contribute to the predicted control command. We conduct comprehensive ablation studies on the feature extraction step and validate our method on the CARLA benchmarks. We also demonstrate that our approach improves interpretability, which correlates with reducing infractions, yielding a safer, high-performance driving model. Notably, our monocular, non-ensemble model surpasses the top-performing approaches from the CARLA Leaderboard by achieving lower infraction scores and the highest route completion rate, all while ensuring interpretability.
Authors: Zixing Wang, Yuhang Zhao
Abstract: End-to-end object detectors offer a promising NMS-free paradigm for real-time applications, yet their high computational cost remains a significant barrier, particularly for complex scenarios like intersection traffic monitoring. To address this challenge, we propose FlowDet, a high-speed detector featuring a decoupled encoder optimization strategy applied to the DETR architecture. Specifically, FlowDet employs a novel Geometric Deformable Unit (GDU) for traffic-aware geometric modeling and a Scale-Aware Attention (SAA) module to maintain high representational power across extreme scale variations. To rigorously evaluate the model's performance in environments with severe occlusion and high object density, we collected the Intersection-Flow-5k dataset, a new challenging scene for this task. Evaluated on Intersection-Flow-5k, FlowDet establishes a new state-of-the-art. Compared to the strong RT-DETR baseline, it improves AP(test) by 1.5% and AP50(test) by 1.6%, while simultaneously reducing GFLOPs by 63.2% and increasing inference speed by 16.2%. Our work demonstrates a new path towards building highly efficient and accurate detectors for demanding, real-world perception systems. The Intersection-Flow-5k dataset is available at https://github.com/AstronZh/Intersection-Flow-5K.
Authors: Antoine Legrand, Renaud Detry, Christophe De Vleeschouwer
Abstract: On-orbit operations require the estimation of the relative 6D pose, i.e., position and orientation, between a chaser spacecraft and its target. While data-driven spacecraft pose estimation methods have been developed, their adoption in real missions is hampered by the lack of understanding of their decision process. This paper presents a method to visualize the 3D visual cues on which a given pose estimator relies. For this purpose, we train a NeRF-based image generator using the gradients back-propagated through the pose estimation network. This enforces the generator to render the main 3D features exploited by the spacecraft pose estimation network. Experiments demonstrate that our method recovers the relevant 3D cues. Furthermore, they offer additional insights on the relationship between the pose estimation network supervision and its implicit representation of the target spacecraft.
Authors: Quanzhu Niu, Dengxian Gong, Shihao Chen, Tao Zhang, Yikang Zhou, Haobo Yuan, Lu Qi, Xiangtai Li, Shunping Ji
Abstract: Referring video object segmentation (RVOS) requires segmenting and tracking objects in videos conditioned on natural-language expressions, demanding fine-grained understanding of both appearance and motion. Building on Sa2VA, which couples a Multi-modal Large Language Model (MLLM) with the video segmentation model SAM2, we identify two key bottlenecks that limit segmentation performance: sparse frame sampling and reliance on a single [SEG] token for an entire video. We propose Segmentation Augmented and Selective Averaged Sa2VA (SaSaSa2VA) to address these issues. On the 7th LSVOS Challenge (RVOS track), SaSaSa2VA achieves a $\mathcal{J\&F}$ of 67.45, ranking first and surpassing the runner-up by 2.80 points. This result and ablation studies demonstrate that efficient segmentation augmentation and test-time ensembling substantially enhance grounded MLLMs for RVOS. The code is released in Sa2VA repository: https://github.com/bytedance/Sa2VA.
Authors: Aniello Panariello, Daniel Marczak, Simone Magistri, Angelo Porrello, Bart{\l}omiej Twardowski, Andrew D. Bagdanov, Simone Calderara, Joost van de Weijer
Abstract: In this paper, we address the challenges associated with merging low-rank adaptations of large neural networks. With the rise of parameter-efficient adaptation techniques, such as Low-Rank Adaptation (LoRA), model fine-tuning has become more accessible. While fine-tuning models with LoRA is highly efficient, existing merging methods often sacrifice this efficiency by merging fully-sized weight matrices. We propose the Core Space merging framework, which enables the merging of LoRA-adapted models within a common alignment basis, thereby preserving the efficiency of low-rank adaptation while substantially improving accuracy across tasks. We further provide a formal proof that projection into Core Space ensures no loss of information and provide a complexity analysis showing the efficiency gains. Extensive empirical results demonstrate that Core Space significantly improves existing merging techniques and achieves state-of-the-art results on both vision and language tasks while utilizing a fraction of the computational resources. Codebase is available at https://github.com/apanariello4/core-space-merging.
Authors: Taohan Weng, Kaibing Hu, Henan Liu, Siya Liu, Xiaoyang Liu, Zhenyu Liu, Jiren Ren, Boyan Wang, Boyang Wang, Yiyu Wang, Yalun Wu, Chaoran Yan, Kaiwen Yan, Jinze Yu, Chi Zhang, Duo Zhang, Haoyun Zheng, Xiaoqing Guo, Jacques Souquet, Hongcheng Guo, Anjie Le
Abstract: Ultrasound is crucial in modern medicine but faces challenges like operator dependence, image noise, and real-time scanning, hindering AI integration. While large multimodal models excel in other medical imaging areas, they struggle with ultrasound's complexities. To address this, we introduce Dolphin v1.0 (V1) and its reasoning-augmented version, Dolphin R1-the first large-scale multimodal ultrasound foundation models unifying diverse clinical tasks in a single vision-language framework.To tackle ultrasound variability and noise, we curated a 2-million-scale multimodal dataset, combining textbook knowledge, public data, synthetic samples, and general corpora. This ensures robust perception, generalization, and clinical adaptability.The Dolphin series employs a three-stage training strategy: domain-specialized pretraining, instruction-driven alignment, and reinforcement-based refinement. Dolphin v1.0 delivers reliable performance in classification, detection, regression, and report generation. Dolphin R1 enhances diagnostic inference, reasoning transparency, and interpretability through reinforcement learning with ultrasound-specific rewards.Evaluated on U2-Bench across eight ultrasound tasks, Dolphin R1 achieves a U2-score of 0.5835-over twice the second-best model (0.2968) setting a new state of the art. Dolphin v1.0 also performs competitively, validating the unified framework. Comparisons show reasoning-enhanced training significantly improves diagnostic accuracy, consistency, and interpretability, highlighting its importance for high-stakes medical AI.
Authors: Yuqing Wang, Zhaiyu Chen, Xiao Xiang Zhu
Abstract: 3D shape completion methods typically assume scans are pre-aligned to a canonical frame. This leaks pose and scale cues that networks may exploit to memorize absolute positions rather than inferring intrinsic geometry. When such alignment is absent in real data, performance collapses. We argue that robust generalization demands architectural equivariance to the similarity group, SIM(3), so the model remains agnostic to pose and scale. Following this principle, we introduce the first SIM(3)-equivariant shape completion network, whose modular layers successively canonicalize features, reason over similarity-invariant geometry, and restore the original frame. Under a de-biased evaluation protocol that removes the hidden cues, our model outperforms both equivariant and augmentation baselines on the PCN benchmark. It also sets new cross-domain records on real driving and indoor scans, lowering minimal matching distance on KITTI by 17% and Chamfer distance $\ell1$ on OmniObject3D by 14%. Perhaps surprisingly, ours under the stricter protocol still outperforms competitors under their biased settings. These results establish full SIM(3) equivariance as an effective route to truly generalizable shape completion. Project page: https://sime-completion.github.io.
Authors: Junbao Zhou, Yuan Zhou, Kesen Zhao, Qingshan Xu, Beier Zhu, Richang Hong, Hanwang Zhang
Abstract: Achieving streaming, fine-grained control over the outputs of autoregressive video diffusion models remains challenging, making it difficult to ensure that they consistently align with user expectations. To bridge this gap, we propose \textbf{stReaming drag-oriEnted interactiVe vidEo manipuLation (REVEL)}, a new task that enables users to modify generated videos \emph{anytime} on \emph{anything} via fine-grained, interactive drag. Beyond DragVideo and SG-I2V, REVEL unifies drag-style video manipulation as editing and animating video frames with both supporting user-specified translation, deformation, and rotation effects, making drag operations versatile. In resolving REVEL, we observe: \emph{i}) drag-induced perturbations accumulate in latent space, causing severe latent distribution drift that halts the drag process; \emph{ii}) streaming drag is easily disturbed by context frames, thereby yielding visually unnatural outcomes. We thus propose a training-free approach, \textbf{DragStream}, comprising: \emph{i}) an adaptive distribution self-rectification strategy that leverages neighboring frames' statistics to effectively constrain the drift of latent embeddings; \emph{ii}) a spatial-frequency selective optimization mechanism, allowing the model to fully exploit contextual information while mitigating its interference via selectively propagating visual cues along generation. Our method can be seamlessly integrated into existing autoregressive video diffusion models, and extensive experiments firmly demonstrate the effectiveness of our DragStream.
Authors: I. M. De la Jara, C. Rodriguez-Opazo, D. Teney, D. Ranasinghe, E. Abbasnejad
Abstract: Out-of-distribution (OOD) detection is essential for reliably deploying machine learning models in the wild. Yet, most methods treat large pre-trained models as monolithic encoders and rely solely on their final-layer representations for detection. We challenge this wisdom. We reveal the \textit{intermediate layers} of pre-trained models, shaped by residual connections that subtly transform input projections, \textit{can} encode \textit{surprisingly rich and diverse signals} for detecting distributional shifts. Importantly, to exploit latent representation diversity across layers, we introduce an entropy-based criterion to \textit{automatically} identify layers offering the most complementary information in a training-free setting -- \textit{without access to OOD data}. We show that selectively incorporating these intermediate representations can increase the accuracy of OOD detection by up to \textbf{$10\%$} in far-OOD and over \textbf{$7\%$} in near-OOD benchmarks compared to state-of-the-art training-free methods across various model architectures and training objectives. Our findings reveal a new avenue for OOD detection research and uncover the impact of various training objectives and model architectures on confidence-based OOD detection methods.
Authors: Ron Keuth, Paul Kaftan, Mattias P. Heinrich
Abstract: The generalization of the Transformer architecture via MetaFormer has reshaped our understanding of its success in computer vision. By replacing self-attention with simpler token mixers, MetaFormer provides strong baselines for vision tasks. However, while extensively studied on natural image datasets, its use in medical imaging remains scarce, and existing works rarely compare different token mixers, potentially overlooking more suitable designs choices. In this work, we present the first comprehensive study of token mixers for medical imaging. We systematically analyze pooling-, convolution-, and attention-based token mixers within the MetaFormer architecture on image classification (global prediction task) and semantic segmentation (dense prediction task). Our evaluation spans eight datasets covering diverse modalities and common challenges in the medical domain. Given the prevalence of pretraining from natural images to mitigate medical data scarcity, we also examine transferring pretrained weights to new token mixers. Our results show that, for classification, low-complexity token mixers (e.g. grouped convolution or pooling) are sufficient, aligning with findings on natural images. Pretrained weights remain useful despite the domain gap introduced by the new token mixer. For segmentation, we find that the local inductive bias of convolutional token mixers is essential. Grouped convolutions emerge as the preferred choice, as they reduce runtime and parameter count compared to standard convolutions, while the MetaFormer's channel-MLPs already provide the necessary cross-channel interactions.
Authors: Xiaoxiao Ma, Feng Zhao, Pengyang Ling, Haibo Qiu, Zhixiang Wei, Hu Yu, Jie Huang, Zhixiong Zeng, Lin Ma
Abstract: In this work, we first revisit the sampling issues in current autoregressive (AR) image generation models and identify that image tokens, unlike text tokens, exhibit lower information density and non-uniform spatial distribution. Accordingly, we present an entropy-informed decoding strategy that facilitates higher autoregressive generation quality with faster synthesis speed. Specifically, the proposed method introduces two main innovations: 1) dynamic temperature control guided by spatial entropy of token distributions, enhancing the balance between content diversity, alignment accuracy, and structural coherence in both mask-based and scale-wise models, without extra computational overhead, and 2) entropy-aware acceptance rules in speculative decoding, achieving near-lossless generation at about 85\% of the inference cost of conventional acceleration methods. Extensive experiments across multiple benchmarks using diverse AR image generation models demonstrate the effectiveness and generalizability of our approach in enhancing both generation quality and sampling speed.
Authors: Dominik Winter, Mai Bui, Monica Azqueta Gavaldon, Nicolas Triltsch, Marco Rosati, Nicolas Brieu
Abstract: Scarcity of annotated data, particularly for rare or atypical morphologies, present significant challenges for cell and nuclei segmentation in computational pathology. While manual annotation is labor-intensive and costly, synthetic data offers a cost-effective alternative. We introduce a Multimodal Semantic Diffusion Model (MSDM) for generating realistic pixel-precise image-mask pairs for cell and nuclei segmentation. By conditioning the generative process with cellular/nuclear morphologies (using horizontal and vertical maps), RGB color characteristics, and BERT-encoded assay/indication metadata, MSDM generates datasests with desired morphological properties. These heterogeneous modalities are integrated via multi-head cross-attention, enabling fine-grained control over the generated images. Quantitative analysis demonstrates that synthetic images closely match real data, with low Wasserstein distances between embeddings of generated and real images under matching biological conditions. The incorporation of these synthetic samples, exemplified by columnar cells, significantly improves segmentation model accuracy on columnar cells. This strategy systematically enriches data sets, directly targeting model deficiencies. We highlight the effectiveness of multimodal diffusion-based augmentation for advancing the robustness and generalizability of cell and nuclei segmentation models. Thereby, we pave the way for broader application of generative models in computational pathology.
Authors: Wangyu Wu, Xuhang Chen, Zhenhong Chen, Jing-En Jiang, Kim-Fung Tsang, Xiaowei Huang, Fei Ma, Jimin Xiao
Abstract: Cross-Domain Sequential Recommendation (CDSR) plays a crucial role in modern consumer electronics and e-commerce platforms, where users interact with diverse services such as books, movies, and online retail products. These systems must accurately capture both domain-specific and cross-domain behavioral patterns to provide personalized and seamless consumer experiences. To address this challenge, we propose \textbf{TEMA-LLM} (\textit{Tag-Enriched Multi-Attention with Large Language Models}), a practical and effective framework that integrates \textit{Large Language Models (LLMs)} for semantic tag generation and enrichment. Specifically, TEMA-LLM employs LLMs to assign domain-aware prompts and generate descriptive tags from item titles and descriptions. The resulting tag embeddings are fused with item identifiers as well as textual and visual features to construct enhanced item representations. A \textit{Tag-Enriched Multi-Attention} mechanism is then introduced to jointly model user preferences within and across domains, enabling the system to capture complex and evolving consumer interests. Extensive experiments on four large-scale e-commerce datasets demonstrate that TEMA-LLM consistently outperforms state-of-the-art baselines, underscoring the benefits of LLM-based semantic tagging and multi-attention integration for consumer-facing recommendation systems. The proposed approach highlights the potential of LLMs to advance intelligent, user-centric services in the field of consumer electronics.
Authors: Samer Al-Hamadani
Abstract: Object detection traditionally relies on costly manual annotation. We present the first comprehensive cost-effectiveness analysis comparing supervised YOLO and zero-shot vision-language models (Gemini Flash 2.5 and GPT-4). Evaluated on 5,000 stratified COCO images and 500 diverse product images, combined with Total Cost of Ownership modeling, we derive break-even thresholds for architecture selection. Results show supervised YOLO attains 91.2% accuracy versus 68.5% for Gemini and 71.3% for GPT-4 on standard categories; the annotation expense for a 100-category system is $10,800, and the accuracy advantage only pays off beyond 55 million inferences (151,000 images/day for one year). On diverse product categories Gemini achieves 52.3% and GPT-4 55.1%, while supervised YOLO cannot detect untrained classes. Cost-per-correct-detection favors Gemini ($0.00050) and GPT-4 ($0.00067) over YOLO ($0.143) at 100,000 inferences. We provide decision frameworks showing that optimal architecture choice depends on inference volume, category stability, budget, and accuracy requirements.
Authors: Wenyuan Zhang, Jimin Tang, Weiqi Zhang, Yi Fang, Yu-Shen Liu, Zhizhong Han
Abstract: Modeling reflections from 2D images is essential for photorealistic rendering and novel view synthesis. Recent approaches enhance Gaussian primitives with reflection-related material attributes to enable physically based rendering (PBR) with Gaussian Splatting. However, the material inference often lacks sufficient constraints, especially under limited environment modeling, resulting in illumination aliasing and reduced generalization. In this work, we revisit the problem from a multi-view perspective and show that multi-view consistent material inference with more physically-based environment modeling is key to learning accurate reflections with Gaussian Splatting. To this end, we enforce 2D Gaussians to produce multi-view consistent material maps during deferred shading. We also track photometric variations across views to identify highly reflective regions, which serve as strong priors for reflection strength terms. To handle indirect illumination caused by inter-object occlusions, we further introduce an environment modeling strategy through ray tracing with 2DGS, enabling photorealistic rendering of indirect radiance. Experiments on widely used benchmarks show that our method faithfully recovers both illumination and geometry, achieving state-of-the-art rendering quality in novel views synthesis.
Authors: Ali Kashefi, Tapan Mukerji
Abstract: Vision Mamba has recently received attention as an alternative to Vision Transformers (ViTs) for image classification. The network size of Vision Mamba scales linearly with input image resolution, whereas ViTs scale quadratically, a feature that improves computational and memory efficiency. Moreover, Vision Mamba requires a significantly smaller number of trainable parameters than traditional convolutional neural networks (CNNs), and thus, they can be more memory efficient. Because of these features, we introduce, for the first time, a neural network that uses Vision Mamba as its backbone for predicting the permeability of three-dimensional porous media. We compare the performance of Vision Mamba with ViT and CNN models across multiple aspects of permeability prediction and perform an ablation study to assess the effects of its components on accuracy. We demonstrate in practice the aforementioned advantages of Vision Mamba over ViTs and CNNs in the permeability prediction of three-dimensional porous media. We make the source code publicly available to facilitate reproducibility and to enable other researchers to build on and extend this work. We believe the proposed framework has the potential to be integrated into large vision models in which Vision Mamba is used instead of ViTs.
Authors: Zhifei Chen, Tianshuo Xu, Leyi Wu, Luozhou Wang, Dongyu Yan, Zihan You, Wenting Luo, Guo Zhang, Yingcong Chen
Abstract: Video generation has recently made striking visual progress, but maintaining coherent object motion and interactions remains difficult. We trace two practical bottlenecks: (i) human-provided motion hints (e.g., small 2D maps) often collapse to too few effective tokens after encoding, weakening guidance; and (ii) optimizing for appearance and motion in a single head can favor texture over temporal consistency. We present STANCE, an image-to-video framework that addresses both issues with two simple components. First, we introduce Instance Cues -- a pixel-aligned control signal that turns sparse, user-editable hints into a dense 2.5D (camera-relative) motion field by averaging per-instance flow and augmenting with monocular depth over the instance mask. This reduces depth ambiguity compared to 2D arrow inputs while remaining easy to use. Second, we preserve the salience of these cues in token space with Dense RoPE, which tags a small set of motion tokens (anchored on the first frame) with spatial-addressable rotary embeddings. Paired with joint RGB \(+\) auxiliary-map prediction (segmentation or depth), our model anchors structure while RGB handles appearance, stabilizing optimization and improving temporal coherence without requiring per-frame trajectory scripts.
Authors: Yuyang Hong, Jiaqi Gu, Qi Yang, Lubin Fan, Yue Wu, Ying Wang, Kun Ding, Shiming Xiang, Jieping Ye
Abstract: Knowledge-based visual question answering (KB-VQA) requires visual language models (VLMs) to integrate visual understanding with external knowledge retrieval. Although retrieval-augmented generation (RAG) achieves significant advances in this task by combining knowledge-base querying, it still struggles with the quality of multimodal queries and the relevance of retrieved results. To overcome these challenges, we propose a novel three-stage method, termed Wiki-PRF, including Processing, Retrieval and Filtering stages. The processing stage dynamically invokes visual tools to extract precise multimodal information for retrieval. The retrieval stage integrates visual and text features to achieve multimodal knowledge retrieval. The filtering stage performs relevance filtering and concentration on retrieval results. To this end, we introduce a visual language model trained with answer accuracy and format consistency as reward signals via a reinforcement learning manner. This enhances the model's reasoning, tool invocation for accurate queries, and filtering of irrelevant content. Experiments on benchmark datasets (E-VQA and InfoSeek) show significant improvements~(36.0 and 42.8) in answer quality, achieving state-of-the-art performance. Code is available at https://github.com/cqu-student/Wiki-PRF
Authors: Daniela Vega, Hannah V. Ceballos, Javier S. Vera, Santiago Rodriguez, Alejandra Perez, Angela Castillo, Maria Escobar, Dario Londo\~no, Luis A. Sarmiento, Camila I. Castro, Nadiezhda Rodriguez, Juan C. Brice\~no, Pablo Arbel\'aez
Abstract: Prenatal diagnosis of Congenital Heart Diseases (CHDs) holds great potential for Artificial Intelligence (AI)-driven solutions. However, collecting high-quality diagnostic data remains difficult due to the rarity of these conditions, resulting in imbalanced and low-quality datasets that hinder model performance. Moreover, no public efforts have been made to integrate multiple sources of information, such as imaging and clinical data, further limiting the ability of AI models to support and enhance clinical decision-making. To overcome these challenges, we introduce the Congenital Anomaly Recognition with Diagnostic Images and Unified Medical records (CARDIUM) dataset, the first publicly available multimodal dataset consolidating fetal ultrasound and echocardiographic images along with maternal clinical records for prenatal CHD detection. Furthermore, we propose a robust multimodal transformer architecture that incorporates a cross-attention mechanism to fuse feature representations from image and tabular data, improving CHD detection by 11% and 50% over image and tabular single-modality approaches, respectively, and achieving an F1 score of 79.8 $\pm$ 4.8% in the CARDIUM dataset. We will publicly release our dataset and code to encourage further research on this unexplored field. Our dataset and code are available at https://github.com/BCV-Uniandes/Cardium, and at the project website https://bcv-uniandes.github.io/CardiumPage/
URLs: https://github.com/BCV-Uniandes/Cardium,, https://bcv-uniandes.github.io/CardiumPage/
Authors: Minglei Shi, Haolin Wang, Wenzhao Zheng, Ziyang Yuan, Xiaoshi Wu, Xintao Wang, Pengfei Wan, Jie Zhou, Jiwen Lu
Abstract: Recent progress in diffusion-based visual generation has largely relied on latent diffusion models with variational autoencoders (VAEs). While effective for high-fidelity synthesis, this VAE+diffusion paradigm suffers from limited training efficiency, slow inference, and poor transferability to broader vision tasks. These issues stem from a key limitation of VAE latent spaces: the lack of clear semantic separation and strong discriminative structure. Our analysis confirms that these properties are crucial not only for perception and understanding tasks, but also for the stable and efficient training of latent diffusion models. Motivated by this insight, we introduce SVG, a novel latent diffusion model without variational autoencoders, which leverages self-supervised representations for visual generation. SVG constructs a feature space with clear semantic discriminability by leveraging frozen DINO features, while a lightweight residual branch captures fine-grained details for high-fidelity reconstruction. Diffusion models are trained directly on this semantically structured latent space to facilitate more efficient learning. As a result, SVG enables accelerated diffusion training, supports few-step sampling, and improves generative quality. Experimental results further show that SVG preserves the semantic and discriminative capabilities of the underlying self-supervised representations, providing a principled pathway toward task-general, high-quality visual representations. Code and interpretations are available at https://howlin-wang.github.io/svg/.
Authors: Shuang Liang, Zhihao Xu, Jialing Tao, Hui Xue, Xiting Wang
Abstract: Despite extensive alignment efforts, Large Vision-Language Models (LVLMs) remain vulnerable to jailbreak attacks, posing serious safety risks. To address this, existing detection methods either learn attack-specific parameters, which hinders generalization to unseen attacks, or rely on heuristically sound principles, which limit accuracy and efficiency. To overcome these limitations, we propose Learning to Detect (LoD), a general framework that accurately detects unknown jailbreak attacks by shifting the focus from attack-specific learning to task-specific learning. This framework includes a Multi-modal Safety Concept Activation Vector module for safety-oriented representation learning and a Safety Pattern Auto-Encoder module for unsupervised attack classification. Extensive experiments show that our method achieves consistently higher detection AUROC on diverse unknown attacks while improving efficiency. The code is available at https://anonymous.4open.science/r/Learning-to-Detect-51CB.
URLs: https://anonymous.4open.science/r/Learning-to-Detect-51CB.
Authors: Xin Li
Abstract: We introduce the delta-homology model of memory, a unified framework in which recall, learning, and prediction emerge from cycle closure, the completion of topologically constrained trajectories within the brain's latent manifold. A Dirac-like memory trace corresponds to a nontrivial homology generator, representing a sparse, irreducible attractor that reactivates only when inference trajectories close upon themselves. In this view, memory is not a static attractor landscape but a topological process of recurrence, where structure arises through the stabilization of closed loops. Building on this principle, we represent spike-timing dynamics as spatiotemporal complexes, in which temporally consistent transitions among neurons form chain complexes supporting persistent activation cycles. These cycles are organized into cell posets, compact causal representations that encode overlapping and compositional memory traces. Within this construction, learning and recall correspond to cycle closure under contextual modulation: inference trajectories stabilize into nontrivial homology classes when both local synchrony (context) and global recurrence (content) are satisfied. We formalize this mechanism through the Context-Content Uncertainty Principle (CCUP), which states that cognition minimizes joint uncertainty between a high-entropy context variable and a low-entropy content variable. Synchronization acts as a context filter selecting coherent subnetworks, while recurrence acts as a content filter validating nontrivial cycles.
Authors: Samiul Based Shuvo, Tasnia Binte Mamun
Abstract: Lung cancer is a leading cause of cancer-related deaths worldwide, and early detection is crucial for improving patient outcomes. Nevertheless, early diagnosis of cancer is a major challenge, particularly in low-resource settings where access to medical resources and trained radiologists is limited. The objective of this study is to propose an automated end-to-end deep learning-based framework for the early detection and classification of lung nodules, specifically for low-resource settings. The proposed framework consists of three stages: lung segmentation using a modified 3D U-Net named 3D Res-U-Net, nodule detection using YOLO-v5, and classification with a Vision Transformer-based architecture. We evaluated the proposed framework on a publicly available dataset, LUNA16. The proposed framework's performance was measured using the respective domain's evaluation matrices. The proposed framework achieved a 98.82% lung segmentation dice score while detecting the lung nodule with 0.76 mAP@50 from the segmented lung, at a low false-positive rate. The performance of both networks of the proposed framework was compared with other studies and found to outperform them regarding segmentation and detection accuracy. Additionally, our proposed Vision transformer network obtained an accuracy of 93.57%, which is 1.21% higher than the state-of-the-art networks. Our proposed end-to-end deep learning-based framework can effectively segment lungs, and detect and classify lung nodules, specifically in low-resource settings with limited access to radiologists. The proposed framework outperforms existing studies regarding all the respective evaluation metrics. The proposed framework can potentially improve the accuracy and efficiency of lung cancer screening in low-resource settings, ultimately leading to better patient outcomes.
Authors: Mohammad Dehghani, Mohadeseh Zarei Ghobadi, Mobin Mohammadi, Diyana Tehrany Dehkordy
Abstract: Objective: Identifying patients at high risk of mortality is crucial for emergency physicians to allocate hospital resources effectively, particularly in regions with limited medical services. This need becomes even more pressing during global health crises that lead to significant morbidity and mortality. This study aimed to present the usability deep neural decision forest and deep neural decision tree to predict mortality among Coronavirus disease 2019 (COVID-19) patients. To this end, We used patient data encompassing Coronavirus disease 2019 diagnosis, demographics, health indicators, and occupational risk factors to analyze disease severity and outcomes. The dataset was partitioned using a stratified sampling method, ensuring that 80% was allocated for training and 20% for testing. Nine machine learning and deep learning methods were employed to build predictive models. The models were evaluated across all stages to determine their effectiveness in predicting patient outcomes. Results: Among the models, the deep neural decision forest consistently outperformed others. Results indicated that using only clinical data yielded an accuracy of 80% by deep neural decision forest, demonstrating it as a reliable predictor of patient mortality. Moreover, the results suggest that clinical data alone may be the most accurate diagnostic tool for predicting mortality.
Authors: Ziang Wu, Jinwei Xie, Xuanyu Zhang, Tao Wang, Yongjun Zhang, Qi Zhu, Chunwei Tian
Abstract: Convolutional neural networks can automatically learn features via deep network architectures and given input samples. However, the robustness of obtained models may face challenges in varying scenes. Bigger differences in network architecture are beneficial to extract more diversified structural information to strengthen the robustness of an obtained super-resolution model. In this paper, we proposed a adaptive convolutional neural network for image super-resolution (ADSRNet). To capture more information, ADSRNet is implemented by a heterogeneous parallel network. The upper network can enhance relation of context information, salient information relation of a kernel mapping and relations of shallow and deep layers to improve performance of image super-resolution. That can strengthen adaptability of an obtained super-resolution model for different scenes. The lower network utilizes a symmetric architecture to enhance relations of different layers to mine more structural information, which is complementary with a upper network for image super-resolution. The relevant experimental results show that the proposed ADSRNet is effective to deal with image resolving. Codes are obtained at https://github.com/hellloxiaotian/ADSRNet.
Authors: Jihoon Cho, Jonghye Woo, Jinah Park
Abstract: Multisequence Magnetic Resonance Imaging (MRI) provides a more reliable diagnosis in clinical applications through complementary information across sequences. However, in practice, the absence of certain MR sequences is a common problem that can lead to inconsistent analysis results. In this work, we propose a novel unified framework for synthesizing multisequence MR images, called hybrid-fusion GAN (HF-GAN). The fundamental mechanism of this work is principled feature disentanglement, which aligns the design of the architecture with the complexity of the features. A powerful many-to-one stream is constructed for the extraction of complex complementary features, while utilizing parallel, one-to-one streams to process modality-specific information. These disentangled features are dynamically integrated into a common latent space by a channel attention-based fusion module (CAFF) and then transformed via a modality infuser to generate the target sequence. We validated our framework on public datasets of both healthy and pathological brain MRI. Quantitative and qualitative results show that HF-GAN achieves state-of-the-art performance, with our 2D slice-based framework notably outperforming a leading 3D volumetric model. Furthermore, the utilization of HF-GAN for data imputation substantially improves the performance of the downstream brain tumor segmentation task, demonstrating its clinical relevance.
Authors: Yonatan Urman, Zachary Shah, Ashwin Kumar, Bruno P. Soares, Kawin Setsompop
Abstract: Purpose: To accelerate MRI acquisition by incorporating the previous scans of a subject during reconstruction. Although longitudinal imaging constitutes much of clinical MRI, leveraging previous scans is challenging due to the complex relationship between scan sessions, potentially involving substantial anatomical or pathological changes, and the lack of open-access datasets with both longitudinal pairs and raw k-space needed for training deep learning-based reconstruction models. Methods: We propose a diffusion-model-based reconstruction framework that eliminates the need for longitudinally paired training data. During training, we treat all scan timepoints as samples from the same distribution, therefore requiring only standalone images. At inference, our framework integrates a subject's prior scan in magnitude DICOM format, which is readily available in clinical workflows, to guide reconstruction of the follow-up. To support future development, we introduce an open-access clinical dataset containing multi-session pairs including prior DICOMs and follow-up k-space. Results: Our method consistently outperforms both longitudinal and non-longitudinal baseline reconstruction methods across various accelerated Cartesian acquisition strategies. In imaging regions highly similar to the prior scan, we observe up to 10\% higher SSIM and 2 dB higher PSNR, without degradation in dissimilar areas. Compared to longitudinal reconstruction baselines, our method demonstrates robustness to varying degrees of anatomical change and misregistration. Conclusion: We demonstrate that prior scans can be effectively integrated with state-of-the-art diffusion-based reconstruction methods to improve image quality and enable greater scan acceleration, without requiring an extensive longitudinally-paired training dataset.
Authors: Kai-liang Lu
Abstract: In many scientific and engineering (e.g., physical, biochemical, medical) practices, data generated through expensive experiments or large-scale simulations, are often sparse and noisy. Physics-informed neural network (PINN) incorporates physical information and knowledge into network topology or computational processes as model priors, with the unique advantage of achieving strong generalization with small data. This study aims to investigate the performance characteristics of the soft-constrained PINN method to solving typical linear and nonlinear ordinary differential equations (ODEs) such as primer, Van der Pol and Duffing oscillators, especially the effectiveness, efficiency, and robustness to noise with minimal data. It is verified that the soft-constrained PINN significantly reduces the need for labeled data. With the aid of appropriate collocation points no need to be labeled, it can predict and also extrapolate with minimal data. First-order and second-order ODEs, no matter linear or nonlinear oscillators, require only one and two training data (containing initial values) respectively, just like classical analytic or Runge-Kutta methods, and with equivalent precision and comparable efficiency (fast training in seconds for scalar ODEs). Furthermore, it can conveniently impose a physical law (e.g., conservation of energy) constraint by adding a regularization term to the total loss function, improving the performance to deal with various complexities such as nonlinearity like Duffing. The DeepXDE-based PINN implementation is light code and can be efficiently trained on both GPU and CPU platforms. The mathematical and computational framework of this alternative and feasible PINN method to ODEs, can be easily extended to PDEs, etc., and is becoming a favorable catalyst for the era of Digital Twins.
Authors: Hassan Gharoun, Mohammad Sadegh Khorshidi, Fang Chen, Amir H. Gandomi
Abstract: Reliable uncertainty quantification is critical in high-stakes applications, such as medical diagnosis, where confidently incorrect predictions can erode trust in automated decision-making systems. Traditional uncertainty quantification methods rely on a predefined confidence threshold to classify predictions as confident or uncertain. However, this approach assumes that predictions exceeding the threshold are trustworthy, while those below it are uncertain, without explicitly assessing the correctness of high-confidence predictions. As a result, confidently incorrect predictions may still occur, leading to misleading uncertainty assessments. To address this limitation, this study proposed an uncertainty-aware stacked neural network, which extends conventional uncertainty quantification by learning when predictions should be trusted. The framework consists of a two-tier model: the base model generates predictions with uncertainty estimates, while the meta-model learns to assign a trust flag, distinguishing confidently correct cases from those requiring expert review. The proposed approach is evaluated against the traditional threshold-based method across multiple confidence thresholds and pre-trained architectures using the COVIDx CXR-4 dataset. Results demonstrate that the proposed framework significantly reduces confidently incorrect predictions, offering a more trustworthy and efficient decision-support system for high-stakes domains.
Authors: Haidong Xu, Meishan Zhang, Hao Ju, Zhedong Zheng, Erik Cambria, Min Zhang, Hao Fei
Abstract: Enabling digital humans to express rich emotions has significant applications in dialogue systems, gaming, and other interactive scenarios. While recent advances in talking head synthesis have achieved impressive results in lip synchronization, they tend to overlook the rich and dynamic nature of facial expressions. To fill this critical gap, we introduce an end-to-end text-to-expression model that explicitly focuses on emotional dynamics. Our model learns expressive facial variations in a continuous latent space and generates expressions that are diverse, fluid, and emotionally coherent. To support this task, we introduce EmoAva, a large-scale and high-quality dataset containing 15,000 text-3D expression pairs. Extensive experiments on both existing datasets and EmoAva demonstrate that our method significantly outperforms baselines across multiple evaluation metrics, marking a significant advancement in the field.
Authors: Wenhui Lei, Hanyu Chen, Zitian Zhang, Luyang Luo, Qiong Xiao, Yannian Gu, Peng Gao, Yankai Jiang, Ci Wang, Guangtao Wu, Tongjia Xu, Yingjie Zhang, Pranav Rajpurkar, Xiaofan Zhang, Shaoting Zhang, Zhenning Wang
Abstract: AI-assisted imaging made substantial advances in tumor diagnosis and management. However, a major barrier to developing robust oncology foundation models is the scarcity of large-scale, high-quality annotated datasets, which are limited by privacy restrictions and the high cost of manual labeling. To address this gap, we present PASTA, a pan-tumor radiology foundation model built on PASTA-Gen, a synthetic data framework that generated 30,000 3D CT scans with pixel-level lesion masks and structured reports of tumors across ten organ systems. Leveraging this resource, PASTA achieves state-of-the-art performance on 45 of 46 oncology tasks, including non-contrast CT tumor screening, lesion segmentation, structured reporting, tumor staging, survival prediction, and MRI-modality transfer. To assess clinical applicability, we developed PASTA-AID, a clinical decision support system, and ran a retrospective simulated clinical trial across two scenarios. For pan-tumor screening on plain CT with fixed reading time, PASTA-AID increased radiologists' throughput by 11.1-25.1% and improved sensitivity by 17.0-31.4% and precision by 10.5-24.9%; additionally, in a diagnosis-aid workflow, it reduced segmentation time by up to 78.2% and reporting time by up to 36.5%. Beyond gains in accuracy and efficiency, PASTA-AID narrowed the expertise gap, enabling less-experienced radiologists to approach expert-level performance. Together, this work establishes an end-to-end, synthetic data-driven pipeline spanning data generation, model development, and clinical validation, thereby demonstrating substantial potential for pan-tumor research and clinical translation.
Authors: Fadillah Maani, Numan Saeed, Tausifa Saleem, Zaid Farooq, Hussain Alasmawi, Werner Diehl, Ameera Mohammad, Gareth Waring, Saudabi Valappi, Leanne Bricker, Mohammad Yaqub
Abstract: Foundation models are becoming increasingly effective in the medical domain, offering pre-trained models on large datasets that can be readily adapted for downstream tasks. Despite progress, fetal ultrasound images remain a challenging domain for foundation models due to their inherent complexity, often requiring substantial additional training and facing limitations due to the scarcity of paired multimodal data. To overcome these challenges, here we introduce FetalCLIP, a vision-language foundation model capable of generating universal representation of fetal ultrasound images. FetalCLIP was pre-trained using a multimodal learning approach on a diverse dataset of 210,035 fetal ultrasound images paired with text. This represents the largest paired dataset of its kind used for foundation model development to date. This unique training approach allows FetalCLIP to effectively learn the intricate anatomical features present in fetal ultrasound images, resulting in robust representations that can be used for a variety of downstream applications. In extensive benchmarking across a range of key fetal ultrasound applications, including classification, gestational age estimation, congenital heart defect (CHD) detection, and fetal structure segmentation, FetalCLIP outperformed all baselines while demonstrating remarkable generalizability and strong performance even with limited labeled data. We plan to release the FetalCLIP model publicly for the benefit of the broader scientific community.
Authors: Teng Zhang, Hongxu Jiang, Kuang Gong, Wei Shao
Abstract: Diffusion models generate data by learning to reverse a forward process, where samples are progressively perturbed with Gaussian noise according to a predefined noise schedule. From a geometric perspective, each noise schedule corresponds to a unique trajectory in probability space from the data distribution to a Gaussian prior. However, prior diffusion models rely on empirically chosen schedules that may not be optimal. This inefficiency necessitates many intermediate time steps, resulting in high computational costs during both training and sampling. To address this, we derive a family of geodesic noise schedules corresponding to the shortest paths in probability space under the Fisher-Rao metric. Based on these schedules, we propose Geodesic Diffusion Models (GDMs), which significantly improve training and sampling efficiency by minimizing the energy required to transform between probability distributions. This efficiency further enables sampling to start from an intermediate distribution in conditional image generation, achieving state-of-the-art results with as few as 6 steps. We evaluated GDM on two medical image enhancement tasks: CT image denoising and MRI image super-resolution. Experimental results show that GDM achieved state-of-the-art performance while reducing training time by 20- to 30-fold compared to Denoising Diffusion Probabilistic Models (DDPMs) and 4- to 6-fold compared to Fast-DDPM, and accelerating sampling by 160- to 170-fold and 1.6-fold, respectively. These gains support the use of GDM for efficient model development and real-time clinical applications. Our code is publicly available at: https://github.com/mirthAI/GDM-VE.
Authors: Che Liu, Yingji Zhang, Dong Zhang, Weijie Zhang, Chenggong Gong, Yu Lu, Shilin Zhou, Ziliang Gan, Ziao Wang, Haipang Wu, Ji Liu, Andr\'e Freitas, Qifan Wang, Zenglin Xu, Rongjuncheng Zhang, Yong Dai
Abstract: This work proposes an industry-level omni-modal large language model (LLM) pipeline that integrates auditory, visual, and linguistic modalities to overcome challenges such as limited tri-modal datasets, high computational costs, and complex feature alignments. Our pipeline consists of three main components: First, a modular framework enabling flexible configuration of various encoder-LLM-decoder architectures. Second, a lightweight training strategy that pre-trains audio-language alignment on the state-of-the-art vision-language model Qwen2.5-VL, thus avoiding the costly pre-training of vision-specific modalities. Third, an audio synthesis pipeline that generates high-quality audio-text data from diverse real-world scenarios, supporting applications such as Automatic Speech Recognition and Speech-to-Speech chat. To this end, we introduce an industry-level omni-modal LLM, Nexus. Extensive experiments validate the efficacy of our pipeline, yielding the following key findings:(1) In the visual understanding task, Nexus exhibits superior performance compared with its backbone model - Qwen2.5-VL-7B, validating the efficiency of our training strategy. (2) Within the English Spoken Question-Answering task, the model achieves better accuracy than the same-period competitor (i.e, MiniCPM-o2.6-7B) in the LLaMA Q. benchmark. (3) In our real-world ASR testset, Nexus achieves outstanding performance, indicating its robustness in real scenarios. (4) In the Speech-to-Text Translation task, our model outperforms Qwen2-Audio-Instruct-7B. (5) In the Text-to-Speech task, based on pretrained vocoder (e.g., Fishspeech1.4 or CosyVoice2.0), Nexus is comparable to its backbone vocoder on Seed-TTS benchmark. (6) An in-depth analysis of tri-modal alignment reveals that incorporating the audio modality enhances representational alignment between vision and language.
Authors: Zhenyu Hou, Senming Tan, Zhihao Zhang, Long Xu, Mengke Zhang, Zhaoqi He, Chao Xu, Fei Gao, Yanjun Cao
Abstract: Terrain analysis is critical for the practical ap- plication of ground mobile robots in real-world tasks, espe- cially in outdoor unstructured environments. In this paper, we propose a novel spatial-temporal traversability assessment method, which aims to enable autonomous robots to effectively navigate through complex terrains. Our approach utilizes sparse Gaussian processes (SGP) to extract geometric features (curvature, gradient, elevation, etc.) directly from point cloud scans. These features are then used to construct a high- resolution local traversability map. Then, we design a spatial- temporal Bayesian Gaussian kernel (BGK) inference method to dynamically evaluate traversability scores, integrating historical and real-time data while considering factors such as slope, flatness, gradient, and uncertainty metrics. GPU acceleration is applied in the feature extraction step, and the system achieves real-time performance. Extensive simulation experiments across diverse terrain scenarios demonstrate that our method outper- forms SOTA approaches in both accuracy and computational efficiency. Additionally, we develop an autonomous navigation framework integrated with the traversability map and validate it with a differential driven vehicle in complex outdoor envi- ronments. Our code will be open-source for further research and development by the community, https://github.com/ZJU-FAST-Lab/FSGP_BGK.
Authors: Rohit Menon, Nils Dengler, Sicong Pan, Gokul Krishna Chenchani, Maren Bennewitz
Abstract: For scene understanding in unstructured environments, an accurate and uncertainty-aware metric-semantic mapping is required to enable informed action selection by autonomous systems. Existing mapping methods often suffer from overconfident semantic predictions, and sparse and noisy depth sensing, leading to inconsistent map representations. In this paper, we therefore introduce EvidMTL, a multi-task learning framework that uses evidential heads for depth estimation and semantic segmentation, enabling uncertainty-aware inference from monocular RGB images. To enable uncertainty-calibrated evidential multi-task learning, we propose a novel evidential depth loss function that jointly optimizes the belief strength of the depth prediction in conjunction with evidential segmentation loss. Building on this, we present EvidKimera, an uncertainty-aware semantic surface mapping framework, which uses evidential depth and semantics prediction for improved 3D metric-semantic consistency. We train and evaluate EvidMTL on the NYUDepthV2 and assess its zero-shot performance on ScanNetV2, demonstrating superior uncertainty estimation compared to conventional approaches while maintaining comparable depth estimation and semantic segmentation. In zero-shot mapping tests on ScanNetV2, EvidKimera outperforms Kimera in semantic surface mapping accuracy and consistency, highlighting the benefits of uncertainty-aware mapping and underscoring its potential for real-world robotic applications.
Authors: Xiang Lan, Feng Wu, Kai He, Qinghao Zhao, Shenda Hong, Mengling Feng
Abstract: While recent multimodal large language models (MLLMs) have advanced automated ECG interpretation, they still face two key limitations: (1) insufficient multimodal synergy between time series signals and visual ECG representations, and (2) limited explainability in linking diagnoses to granular waveform evidence. We introduce GEM, the first MLLM unifying ECG time series, 12-lead ECG images and text for grounded and clinician-aligned ECG interpretation. GEM enables feature-grounded analysis, evidence-driven reasoning, and a clinician-like diagnostic process through three core innovations: a dual-encoder framework extracting complementary time series and image features, cross-modal alignment for effective multimodal understanding, and knowledge-guided instruction generation for generating high-granularity grounding data (ECG-Grounding) linking diagnoses to measurable parameters ($e.g.$, QRS/PR Intervals). Additionally, we propose the Grounded ECG Understanding task, a clinically motivated benchmark designed to comprehensively assess the MLLM's capability in grounded ECG understanding. Experimental results on both existing and our proposed benchmarks show GEM significantly improves predictive performance (CSN $7.4\% \uparrow$), explainability ($22.7\% \uparrow$), and grounding ($24.8\% \uparrow$), making it more suitable for real-world clinical applications. GitHub repository: https://github.com/lanxiang1017/GEM.git
Authors: Mohammad Tariqul Islam, Jason W. Fleischer
Abstract: Uniform manifold approximation and projection (UMAP) is among the most popular neighbor embedding methods. The method relies on attractive and repulsive forces among high-dimensional data points to obtain a low-dimensional embedding. In this paper, we analyze the forces to reveal their effects on cluster formations and visualization and compare UMAP to its contemporaries. Repulsion emphasizes differences, controlling cluster boundaries and inter-cluster distance. Attraction is more subtle, as attractive tension between points can manifest simultaneously as attraction and repulsion in the lower-dimensional mapping. This explains the need for learning rate annealing and motivates the different treatments between attractive and repulsive terms. Moreover, by modifying attraction, we improve the consistency of cluster formation under random initialization. Overall, our analysis makes UMAP and similar embedding methods more interpretable, more robust, and more accurate.
Authors: Yu Qiao, Phuong-Nam Tran, Ji Su Yoon, Loc X. Nguyen, Eui-Nam Huh, Dusit Niyato, Choong Seon Hong
Abstract: Reinforcement learning (RL)-based large language models (LLMs), such as ChatGPT, DeepSeek, and Grok-3, have attracted widespread attention for their remarkable capabilities in multimodal data understanding. Meanwhile, the rapid expansion of information services has led to a growing demand for AI-enabled wireless networks. The open-source DeepSeek models are famous for their innovative designs, such as large-scale pure RL and cost-efficient training, which make them well-suited for practical deployment in wireless networks. By integrating DeepSeek-style LLMs with wireless infrastructures, a synergistic opportunity arises: the DeepSeek-style LLMs enhance network optimization with strong reasoning and decision-making abilities, while wireless infrastructure enables the broad deployment of these models. Motivated by this convergence, this survey presents a comprehensive DeepSeek-inspired exploration of RL-based LLMs in the context of wireless networks. We begin by reviewing key techniques behind network optimization to establish a foundation for understanding DeepSeek-style LLM integration. Next, we examine recent advancements in RL-based LLMs, using DeepSeek models as a representative example. Building on this, we explore the synergy between the two domains, highlighting motivations, challenges, and potential solutions. Finally, we highlight emerging directions for integrating LLMs with wireless networks, such as quantum, on-device, and neural-symbolic LLM models, as well as embodied AI agents. Overall, this survey offers a comprehensive examination of the interplay between DeepSeek-style LLMs and wireless networks, demonstrating how these domains can mutually enhance each other to drive innovation.
Authors: Ryan Banks, Vishal Thengane, Mar\'ia Eugenia Guerrero, Nelly Maria Garc\'ia-Madue\~no, Yunpeng Li, Hongying Tang, Akhilanand Chaurasia
Abstract: This study proposes a deep learning framework and annotation methodology for the automatic detection of periodontal bone loss landmarks, associated conditions, and staging. 192 periapical radiographs were collected and annotated with a stage agnostic methodology, labelling clinically relevant landmarks regardless of disease presence or extent. We propose a heuristic post-processing module that aligns predicted keypoints to tooth boundaries using an auxiliary instance segmentation model. An evaluation metric, Percentage of Relative Correct Keypoints (PRCK), is proposed to capture keypoint performance in dental imaging domains. Four donor pose estimation models were adapted with fine-tuning for our keypoint problem. Post-processing improved fine-grained localisation, raising average PRCK^{0.05} by +0.028, but reduced coarse performance for PRCK^{0.25} by -0.0523 and PRCK^{0.5} by -0.0345. Orientation estimation shows excellent performance for auxiliary segmentation when filtered with either stage 1 object detection model. Periodontal staging was detected sufficiently, with the best mesial and distal Dice scores of 0.508 and 0.489, while furcation involvement and widened periodontal ligament space tasks remained challenging due to scarce positive samples. Scalability is implied with similar validation and external set performance. The annotation methodology enables stage agnostic training with balanced representation across disease severities for some detection tasks. The PRCK metric provides a domain-specific alternative to generic pose metrics, while the heuristic post-processing module consistently corrected implausible predictions with occasional catastrophic failures. The proposed framework demonstrates the feasibility of clinically interpretable periodontal bone loss assessment, with potential to reduce diagnostic variability and clinician workload.
Authors: Jiaming Zhou, Ke Ye, Jiayi Liu, Teli Ma, Zifan Wang, Ronghe Qiu, Kun-Yu Lin, Zhilin Zhao, Junwei Liang
Abstract: The generalization capabilities of vision-language-action (VLA) models to unseen tasks are crucial to achieving general-purpose robotic manipulation in open-world settings. However, the cross-task generalization capabilities of existing VLA models remain significantly underexplored. To address this gap, we introduce AGNOSTOS, a novel simulation benchmark designed to rigorously evaluate cross-task zero-shot generalization in manipulation. AGNOSTOS comprises 23 unseen manipulation tasks for testing, distinct from common training task distributions, and incorporates two levels of generalization difficulty to assess robustness. Our systematic evaluation reveals that current VLA models, despite being trained on diverse datasets, struggle to generalize effectively to these unseen tasks. To overcome this limitation, we propose Cross-Task In-Context Manipulation (X-ICM), a method that conditions large language models (LLMs) on in-context demonstrations from seen tasks to predict action sequences for unseen tasks. Additionally, we introduce a dynamics-guided sample selection strategy that identifies relevant demonstrations by capturing cross-task dynamics. On AGNOSTOS, X-ICM significantly improves cross-task zero-shot generalization performance over leading VLAs. We believe AGNOSTOS and X-ICM will serve as valuable tools for advancing general-purpose robotic manipulation.
Authors: Hao Fang, Changle Zhou, Jiawei Kong, Kuofeng Gao, Bin Chen, Tao Liang, Guojun Ma, Shu-Tao Xia
Abstract: Large Vision-Language Models (LVLMs) are susceptible to hallucinations, where generated responses seem semantically plausible yet exhibit little or no relevance to the input image. Previous studies reveal that this issue primarily stems from LVLMs' over-reliance on language priors while disregarding the visual information during decoding. To alleviate this issue, we introduce a novel Conditional Pointwise Mutual Information (C-PMI) calibrated decoding strategy, which adaptively strengthens the mutual dependency between generated texts and input images to mitigate hallucinations. Unlike existing methods solely focusing on text token sampling, we propose to jointly model the contributions of visual and textual tokens to C-PMI, formulating hallucination mitigation as a bi-level optimization problem aimed at maximizing mutual information. To solve it, we design a token purification mechanism that dynamically regulates the decoding process by sampling text tokens remaining maximally relevant to the given image, while simultaneously refining image tokens most pertinent to the generated response. Extensive experiments across various benchmarks reveal that the proposed method significantly reduces hallucinations in LVLMs while preserving decoding efficiency.
Authors: Enshen Zhou, Jingkun An, Cheng Chi, Yi Han, Shanyu Rong, Chi Zhang, Pengwei Wang, Zhongyuan Wang, Tiejun Huang, Lu Sheng, Shanghang Zhang
Abstract: Spatial referring is a fundamental capability of embodied robots to interact with the 3D physical world. However, even with the powerful pretrained vision language models (VLMs), recent approaches are still not qualified to accurately understand the complex 3D scenes and dynamically reason about the instruction-indicated locations for interaction. To this end, we propose RoboRefer, a 3D-aware VLM that can first achieve precise spatial understanding by integrating a disentangled but dedicated depth encoder via supervised fine-tuning (SFT). Moreover, RoboRefer advances generalized multi-step spatial reasoning via reinforcement fine-tuning (RFT), with metric-sensitive process reward functions tailored for spatial referring tasks. To support SFT and RFT training, we introduce RefSpatial, a large-scale dataset of 20M QA pairs (2x prior), covering 31 spatial relations (vs. 15 prior) and supporting complex reasoning processes (up to 5 steps). In addition, we introduce RefSpatial-Bench, a challenging benchmark filling the gap in evaluating spatial referring with multi-step reasoning. Experiments show that SFT-trained RoboRefer achieves state-of-the-art spatial understanding, with an average success rate of 89.6%. RFT-trained RoboRefer further outperforms all other baselines by a large margin, even surpassing Gemini-2.5-Pro by 17.4% in average accuracy on RefSpatial-Bench. Notably, RoboRefer can be integrated with various control policies to execute long-horizon, dynamic tasks across diverse robots (e,g., UR5, G1 humanoid) in cluttered real-world scenes. See the project page at https://zhoues.github.io/RoboRefer.
Authors: Maxence Boels, Harry Robertshaw, Thomas C Booth, Prokar Dasgupta, Alejandro Granados, Sebastien Ourselin
Abstract: Surgical action planning requires predicting future instrument-verb-target triplets for real-time assistance. While teleoperated robotic surgery provides natural expert demonstrations for imitation learning (IL), reinforcement learning (RL) could potentially discover superior strategies through self-exploration. We present the first comprehensive comparison of IL versus RL for surgical action planning on CholecT50. Our Dual-task Autoregressive Imitation Learning (DARIL) baseline achieves 34.6% action triplet recognition mAP and 33.6% next frame prediction mAP with smooth planning degradation to 29.2% at 10-second horizons. We evaluated three RL variants: world model-based RL, direct video RL, and inverse RL enhancement. Surprisingly, all RL approaches underperformed DARIL--world model RL dropped to 3.1% mAP at 10s while direct video RL achieved only 15.9%. Our analysis reveals that distribution matching on expert-annotated test sets systematically favors IL over potentially valid RL policies that differ from training demonstrations. This challenges assumptions about RL superiority in sequential decision making and provides crucial insights for surgical AI development.
Authors: Chun Xie, Yuichi Yoshii, Itaru Kitahara
Abstract: X-ray imaging is a rapid and cost-effective tool for visualizing internal human anatomy. While multi-view X-ray imaging provides complementary information that enhances diagnosis, intervention, and education, acquiring images from multiple angles increases radiation exposure and complicates clinical workflows. To address these challenges, we propose a novel view-conditioned diffusion model for synthesizing multi-view X-ray images from a single view. Unlike prior methods, which are limited in angular range, resolution, and image quality, our approach leverages the Diffusion Transformer to preserve fine details and employs a weak-to-strong training strategy for stable high-resolution image generation. Experimental results demonstrate that our method generates higher-resolution outputs with improved control over viewing angles. This capability has significant implications not only for clinical applications but also for medical education and data extension, enabling the creation of diverse, high-quality datasets for training and analysis. Our code is available at https://github.com/xiechun298/SV-DRR.
Authors: Mingda Zhang
Abstract: Accurate brain tumor segmentation is crucial for neuro-oncology diagnosis and treatment planning. Deep learning methods have made significant progress, but automatic segmentation still faces challenges, including tumor morphological heterogeneity and complex three-dimensional spatial relationships. This paper proposes a three-tier fusion architecture that achieves precise brain tumor segmentation. The method processes information progressively at the pixel, feature, and semantic levels. At the pixel level, physical modeling extends magnetic resonance imaging (MRI) to multimodal data, including simulated ultrasound and synthetic computed tomography (CT). At the feature level, the method performs Transformer-based cross-modal feature fusion through multi-teacher collaborative distillation, integrating three expert teachers (MRI, US, CT). At the semantic level, clinical textual knowledge generated by GPT-4V is transformed into spatial guidance signals using CLIP contrastive learning and Feature-wise Linear Modulation (FiLM). These three tiers together form a complete processing chain from data augmentation to feature extraction to semantic guidance. We validated the method on the Brain Tumor Segmentation (BraTS) 2020, 2021, and 2023 datasets. The model achieves average Dice coefficients of 0.8665, 0.9014, and 0.8912 on the three datasets, respectively, and reduces the 95% Hausdorff Distance (HD95) by an average of 6.57 millimeters compared with the baseline. This method provides a new paradigm for precise tumor segmentation and boundary localization.
Authors: Ritik Shah, Marco F. Duarte
Abstract: High-spatial-resolution hyperspectral images (HSI) are essential for applications such as remote sensing and medical imaging, yet HSI sensors inherently trade spatial detail for spectral richness. Fusing high-spatial-resolution multispectral images (HR-MSI) with low-spatial-resolution hyperspectral images (LR-HSI) is a promising route to recover fine spatial structures without sacrificing spectral fidelity. Most state-of-the-art methods for HSI-MSI fusion demand point spread function (PSF) calibration or ground truth high resolution HSI (HR-HSI), both of which are impractical to obtain in real world settings. We present SpectraLift, a fully self-supervised framework that fuses LR-HSI and HR-MSI inputs using only the MSI's Spectral Response Function (SRF). SpectraLift trains a lightweight per-pixel multi-layer perceptron (MLP) network using ($i$)~a synthetic low-spatial-resolution multispectral image (LR-MSI) obtained by applying the SRF to the LR-HSI as input, ($ii$)~the LR-HSI as the output, and ($iii$)~an $\ell_1$ spectral reconstruction loss between the estimated and true LR-HSI as the optimization objective. At inference, SpectraLift uses the trained network to map the HR-MSI pixel-wise into a HR-HSI estimate. SpectraLift converges in minutes, is agnostic to spatial blur and resolution, and outperforms state-of-the-art methods on PSNR, SAM, SSIM, and RMSE benchmarks.
Authors: Hongzhao Chen, Hexiao Ding, Yufeng Jiang, Jing Lan, Ka Chun Li, Gerald W. Y. Cheng, Nga-Chun Ng, Yao Pu, Jing Cai, Liang-ting Lin, Jung Sun Yoo
Abstract: Reliable and interpretable tumor classification from clinical imaging remains a core challenge. The main difficulties arise from heterogeneous modality quality, limited annotations, and the absence of structured anatomical guidance. We present REACT-KD, a Region-Aware Cross-modal Topological Knowledge Distillation framework that transfers supervision from high-fidelity multi-modal sources into a lightweight CT-based student model. The framework employs a dual teacher design. One branch captures structure-function relationships through dual-tracer PET/CT, while the other models dose-aware features using synthetically degraded low-dose CT. These branches jointly guide the student model through two complementary objectives. The first achieves semantic alignment through logits distillation, and the second models anatomical topology through region graph distillation. A shared CBAM3D module ensures consistent attention across modalities. To improve reliability in deployment, REACT-KD introduces modality dropout during training, which enables robust inference under partial or noisy inputs. As a case study, we applied REACT-KD to hepatocellular carcinoma staging. The framework achieved an average AUC of 93.5\% on an internal PET/CT cohort and maintained 76.6\% to 81.5\% AUC across varying levels of dose degradation in external CT testing. Decision curve analysis further shows that REACT-KD consistently provides the highest net clinical benefit across all thresholds, confirming its value in real-world diagnostic practice. Code is available at: https://github.com/Kinetics-JOJO/REACT-KD
Authors: Daniil Zverev, Thadd\"aus Wiedemer, Ameya Prabhu, Matthias Bethge, Wieland Brendel, A. Sophia Koepke
Abstract: The emergence of audio-visual foundation models underscores the importance of reliably assessing their multi-modal understanding. The VGGSound dataset is commonly used as a benchmark for evaluation audio-visual classification. However, our analysis identifies several limitations of VGGSound, including incomplete labelling, partially overlapping classes, and misaligned modalities. These lead to distorted evaluations of auditory and visual capabilities. To address these limitations, we introduce VGGSounder, a comprehensively re-annotated, multi-label test set that extends VGGSound and is specifically designed to evaluate audio-visual foundation models. VGGSounder features detailed modality annotations, enabling precise analyses of modality-specific performance. Furthermore, we reveal model limitations by analysing performance degradation when adding another input modality with our new modality confusion metric.
Authors: Shreya Gummadi, Mateus V. Gasparino, Gianluca Capezzuto, Marcelo Becker, Girish Chowdhary
Abstract: The advancement of robotics and autonomous navigation systems hinges on the ability to accurately predict terrain traversability. Traditional methods for generating datasets to train these prediction models often involve putting robots into potentially hazardous environments, posing risks to equipment and safety. To solve this problem, we present ZeST, a novel approach leveraging visual reasoning capabilities of Large Language Models (LLMs) to create a traversability map in real-time without exposing robots to danger. Our approach not only performs zero-shot traversability and mitigates the risks associated with real-world data collection but also accelerates the development of advanced navigation systems, offering a cost-effective and scalable solution. To support our findings, we present navigation results, in both controlled indoor and unstructured outdoor environments. As shown in the experiments, our method provides safer navigation when compared to other state-of-the-art methods, constantly reaching the final goal.
Authors: Rapha\"el Bourgade, Guillaume Balezo, Hana Feki, Lily Monier, Matthieu Blons, Alice Blondel, Delphine Loussouarn, Anne Vincent-Salomon, Thomas Walter
Abstract: Mitotic figures represent a key histoprognostic feature in tumor pathology, providing crucial insights into tumor aggressiveness and proliferation. However, their identification remains challenging, subject to significant inter-observer variability, even among experienced pathologists. To address this issue, the MItosis DOmain Generalization (MIDOG) 2025 challenge marks the third edition of an international competition aiming to develop robust mitosis detection algorithms. In this paper, we present a mitotic figure detection approach based on the state-of-the-art YOLOv12 object detection architecture. Our method achieved an F1-score of 0.801 on the preliminary test set (hotspots only) and ranked second on the final test leaderboard with an F1-score of 0.7216 across complex and heterogeneous whole-slide regions, without relying on external data.
Authors: Jiong Lin, Jialong Ning, Judah Goldfeder, Hod Lipson
Abstract: In this paper, we formulate the problem of kinematic synthesis for planar linkages as a cross-domain image generation task. We develop a planar linkages dataset using RGB image representations, covering a range of mechanisms: from simple types such as crank-rocker and crank-slider to more complex eight-bar linkages like Jansen's mechanism. A shared-latent variational autoencoder (VAE) is employed to explore the potential of image generative models for synthesizing unseen motion curves and simulating novel kinematics. By encoding the drawing speed of trajectory points as color gradients, the same architecture also supports kinematic synthesis conditioned on both trajectory shape and velocity profiles. We validate our method on three datasets of increasing complexity: a standard four-bar linkage set, a mixed set of four-bar and crank-slider mechanisms, and a complex set including multi-loop mechanisms. Preliminary results demonstrate the effectiveness of image-based representations for generative mechanical design, showing that mechanisms with revolute and prismatic joints, and potentially cams and gears, can be represented and synthesized within a unified image generation framework.
Authors: Yuval Golbari, Navve Wasserman, Gal Vardi, Michal Irani
Abstract: Determining which data samples were used to train a model-known as Membership Inference Attack (MIA)-is a well-studied and important problem with implications for data privacy. Black-box methods presume access only to the model's outputs and often rely on training auxiliary reference models. While they have shown strong empirical performance, they rely on assumptions that rarely hold in real-world settings: (i) the attacker knows the training hyperparameters; (ii) all available non-training samples come from the same distribution as the training data; and (iii) the fraction of training data in the evaluation set is known. In this paper, we demonstrate that removing these assumptions leads to a significant drop in the performance of black-box attacks. We introduce ImpMIA, a Membership Inference Attack that exploits the Implicit Bias of neural networks, hence removes the need to rely on any reference models and their assumptions. ImpMIA is a white-box attack -- a setting which assumes access to model weights and is becoming increasingly realistic given that many models are publicly available (e.g., via Hugging Face). Building on maximum-margin implicit bias theory, ImpMIA uses the Karush-Kuhn-Tucker (KKT) optimality conditions to identify training samples. This is done by finding the samples whose gradients most strongly reconstruct the trained model's parameters. As a result, ImpMIA achieves state-of-the-art performance compared to both black and white box attacks in realistic settings where only the model weights and a superset of the training data are available.
Authors: Chenlong He, Zhijian Hao, Leilei Huang, Xiaoyang Zeng, Yibo Fan
Abstract: Just Noticeable Distortion (JND)-guided pre-filter is a promising technique for improving the perceptual compression efficiency of image coding. However, existing methods are often computationally expensive, and the field lacks standardized benchmarks for fair comparison. To address these challenges, this paper introduces a twofold contribution. First, we develop and open-source FJNDF-Pytorch, a unified benchmark for frequency-domain JND-Guided pre-filters. Second, leveraging this platform, we propose a complete learning framework for a novel, lightweight Convolutional Neural Network (CNN). Experimental results demonstrate that our proposed method achieves state-of-the-art compression efficiency, consistently outperforming competitors across multiple datasets and encoders. In terms of computational cost, our model is exceptionally lightweight, requiring only 7.15 GFLOPs to process a 1080p image, which is merely 14.1% of the cost of recent lightweight network. Our work presents a robust, state-of-the-art solution that excels in both performance and efficiency, supported by a reproducible research platform. The open-source implementation is available at https://github.com/viplab-fudan/FJNDF-Pytorch.
Authors: Wenqian Zhang, Weiyang Liu, Zhen Liu
Abstract: The design of complex machines stands as both a marker of human intelligence and a foundation of engineering practice. Given recent advances in large language models (LLMs), we ask whether they, too, can learn to create. We approach this question through the lens of compositional machine design: a task in which machines are assembled from standardized components to meet functional demands like locomotion or manipulation in a simulated physical environment. With this simplification, machine design is expressed as writing XML-like code that explicitly specifies pairwise part connections. To support this investigation, we introduce BesiegeField, a testbed built on the machine-building game Besiege, which enables part-based construction, physical simulation and reward-driven evaluation. Using BesiegeField, we benchmark state-of-the-art LLMs with agentic workflows and identify key capabilities required for success, including spatial reasoning, strategic assembly, and instruction-following. As current open-source models fall short, we explore reinforcement learning (RL) as a path to improvement: we curate a cold-start dataset, conduct RL finetuning experiments, and highlight open challenges at the intersection of language, machine design, and physical reasoning.