Authors: Karthik Shivashankar
Abstract: The prospect of 4D video in Extended Reality (XR) platform is huge and exciting, it opens a whole new way of human computer interaction and the way we perceive the reality and consume multimedia. In this thesis, we have shown that feasibility of rendering 4D video in Microsoft mixed reality platform. This enables us to port any 3D performance capture from CVSSP into XR product like the HoloLens device with relative ease. However, if the 3D model is too complex and is made up of millions of vertices, the data bandwidth required to port the model is a severe limitation with the current hardware and communication system. Therefore, in this project we have also developed a compact representation of both shape and appearance of the 4d video sequence using deep learning models to effectively learn the compact representation of 4D video sequence and reconstruct it without affecting the shape and appearance of the video sequence.
Authors: Sudong Wang, Yunjian Zhang, Yao Zhu, Jianing Li, Zizhe Wang, Yanwei Liu, Xiangyang Ji
Abstract: Large Vision-Language Models (LVLMs) are gradually becoming the foundation for many artificial intelligence applications. However, understanding their internal working mechanisms has continued to puzzle researchers, which in turn limits the further enhancement of their capabilities. In this paper, we seek to investigate how multimodal knowledge evolves and eventually induces natural languages in LVLMs. We design a series of novel strategies for analyzing internal knowledge within LVLMs, and delve into the evolution of multimodal knowledge from three levels, including single token probabilities, token probability distributions, and feature encodings. In this process, we identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation. Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms. Our codes are available at https://github.com/XIAO4579/Vlm-interpretability.
Authors: Xiucheng Liang, Jinheng Xie, Tianhong Zhao, Rudi Stouffs, Filip Biljecki
Abstract: Building properties, such as height, usage, and material composition, play a crucial role in spatial data infrastructures, supporting applications such as energy simulation, risk assessment, and environmental modeling. Despite their importance, comprehensive and high-quality building attribute data remain scarce in many urban areas. Recent advances have enabled the extraction and tagging of objective building attributes using remote sensing and street-level imagery. However, establishing a method and pipeline that integrates diverse open datasets, acquires holistic building imagery at scale, and infers comprehensive building attributes remains a significant challenge. Among the first, this study bridges the gaps by introducing OpenFACADES, an open framework that leverages multimodal crowdsourced data to enrich building profiles with both objective attributes and semantic descriptors through multimodal large language models. Our methodology proceeds in three major steps. First, we integrate street-level image metadata from Mapillary with OpenStreetMap geometries via isovist analysis, effectively identifying images that provide suitable vantage points for observing target buildings. Second, we automate the detection of building facades in panoramic imagery and tailor a reprojection approach to convert objects into holistic perspective views that approximate real-world observation. Third, we introduce an innovative approach that harnesses and systematically investigates the capabilities of open-source large vision-language models (VLMs) for multi-attribute prediction and open-vocabulary captioning in building-level analytics, leveraging a globally sourced dataset of 30,180 labeled images from seven cities. Evaluation shows that fine-tuned VLM excel in multi-attribute inference, outperforming single-attribute computer vision models and zero-shot ChatGPT-4o.
Authors: Yangxiao Lu, Ruosen Li, Liqiang Jing, Jikai Wang, Xinya Du, Yunhui Guo, Nicholas Ruozzi, Yu Xiang
Abstract: Visual grounding focuses on detecting objects from images based on language expressions. Recent Large Vision-Language Models (LVLMs) have significantly advanced visual grounding performance by training large models with large-scale datasets. However, the problem remains challenging, especially when similar objects appear in the input image. For example, an LVLM may not be able to differentiate Diet Coke and regular Coke in an image. In this case, if additional reference images of Diet Coke and regular Coke are available, it can help the visual grounding of similar objects. In this work, we introduce a new task named Multimodal Reference Visual Grounding (MRVG). In this task, a model has access to a set of reference images of objects in a database. Based on these reference images and a language expression, the model is required to detect a target object from a query image. We first introduce a new dataset to study the MRVG problem. Then we introduce a novel method, named MRVG-Net, to solve this visual grounding problem. We show that by efficiently using reference images with few-shot object detection and using Large Language Models (LLMs) for object matching, our method achieves superior visual grounding performance compared to the state-of-the-art LVLMs such as Qwen2.5-VL-7B. Our approach bridges the gap between few-shot detection and visual grounding, unlocking new capabilities for visual understanding. Project page with our code and dataset: https://irvlutd.github.io/MultiGrounding
Authors: Lilin Xu, Kaiyuan Hou, Xiaofan Jiang
Abstract: Human activity recognition (HAR) using inertial measurement units (IMUs) increasingly leverages large language models (LLMs), yet existing approaches focus on coarse activities like walking or running. Our preliminary study indicates that pretrained LLMs fail catastrophically on fine-grained HAR tasks such as air-written letter recognition, achieving only near-random guessing accuracy. In this work, we first bridge this gap for flat-surface writing scenarios: by fine-tuning LLMs with a self-collected dataset and few-shot learning, we achieved up to a 129x improvement on 2D data. To extend this to 3D scenarios, we designed an encoder-based pipeline that maps 3D data into 2D equivalents, preserving the spatiotemporal information for robust letter prediction. Our end-to-end pipeline achieves 78% accuracy on word recognition with up to 5 letters in mid-air writing scenarios, establishing LLMs as viable tools for fine-grained HAR.
Authors: Baba Ibrahim (Hubei University of Automotive Technology,Hubei University of Automotive Technology), Zhou Kui (Hubei University of Automotive Technology,Hubei University of Automotive Technology)
Abstract: This paper Traffic sign recognition plays a crucial role in the development of autonomous vehicles and advanced driver-assistance systems (ADAS). Despite significant advances in deep learning and object detection, accurately detecting and classifying traffic signs remains challenging due to their small sizes, variable environmental conditions, occlusion, and class imbalance. This thesis presents an enhanced YOLOv8-based detection system that integrates advanced data augmentation techniques, novel architectural enhancements including Coordinate Attention (CA), Bidirectional Feature Pyramid Network (BiFPN), and dynamic modules such as ODConv and LSKA, along with refined loss functions (EIoU and WIoU combined with Focal Loss). Extensive experiments conducted on datasets including GTSRB, TT100K, and GTSDB demonstrate marked improvements in detection accuracy, robustness under adverse conditions, and real-time inference on edge devices. The findings contribute actionable insights for deploying reliable traffic sign recognition systems in real-world autonomous driving scenarios.
Authors: Farida Al Haddad, Yuxin Wang, Malcolm Mielle
Abstract: Artificial intelligence has the potential to impact safety and efficiency in safety-critical domains such as construction, manufacturing, and healthcare. For example, using sensor data from wearable devices, such as inertial measurement units (IMUs), human gestures can be detected while maintaining privacy, thereby ensuring that safety protocols are followed. However, strict safety requirements in these domains have limited the adoption of AI, since accurate calibration of predicted probabilities and robustness against out-of-distribution (OOD) data is necessary. This paper proposes UAC (Uncertainty-Aware Calibration), a novel two-step method to address these challenges in IMU-based gesture recognition. First, we present an uncertainty-aware gesture network architecture that predicts both gesture probabilities and their associated uncertainties from IMU data. This uncertainty is then used to calibrate the probabilities of each potential gesture. Second, an entropy-weighted expectation of predictions over multiple IMU data windows is used to improve accuracy while maintaining correct calibration. Our method is evaluated using three publicly available IMU datasets for gesture detection and is compared to three state-of-the-art calibration methods for neural networks: temperature scaling, entropy maximization, and Laplace approximation. UAC outperforms existing methods, achieving improved accuracy and calibration in both OOD and in-distribution scenarios. Moreover, we find that, unlike our method, none of the state-of-the-art methods significantly improve the calibration of IMU-based gesture recognition models. In conclusion, our work highlights the advantages of uncertainty-aware calibration of neural networks, demonstrating improvements in both calibration and accuracy for gesture detection using IMU data.
Authors: Matheus Martins Batista
Abstract: The growing threat posed by deepfake videos, capable of manipulating realities and disseminating misinformation, drives the urgent need for effective detection methods. This work investigates and compares different approaches for identifying deepfakes, focusing on the GenConViT model and its performance relative to other architectures present in the DeepfakeBenchmark. To contextualize the research, the social and legal impacts of deepfakes are addressed, as well as the technical fundamentals of their creation and detection, including digital image processing, machine learning, and artificial neural networks, with emphasis on Convolutional Neural Networks (CNNs), Generative Adversarial Networks (GANs), and Transformers. The performance evaluation of the models was conducted using relevant metrics and new datasets established in the literature, such as WildDeep-fake and DeepSpeak, aiming to identify the most effective tools in the battle against misinformation and media manipulation. The obtained results indicated that GenConViT, after fine-tuning, exhibited superior performance in terms of accuracy (93.82%) and generalization capacity, surpassing other architectures in the DeepfakeBenchmark on the DeepSpeak dataset. This study contributes to the advancement of deepfake detection techniques, offering contributions to the development of more robust and effective solutions against the dissemination of false information.
Authors: Rohit Agarwal, Aryan Dessai, Arif Ahmed Sekh, Krishna Agarwal, Alexander Horsch, Dilip K. Prasad
Abstract: The field of varying feature space in online learning settings, also known as haphazard inputs, is very prominent nowadays due to its applicability in various fields. However, the current solutions to haphazard inputs are model-dependent and cannot benefit from the existing advanced deep-learning methods, which necessitate inputs of fixed dimensions. Therefore, we propose to transform the varying feature space in an online learning setting to a fixed-dimension image representation on the fly. This simple yet novel approach is model-agnostic, allowing any vision-based models to be applicable for haphazard inputs, as demonstrated using ResNet and ViT. The image representation handles the inconsistent input data seamlessly, making our proposed approach scalable and robust. We show the efficacy of our method on four publicly available datasets. The code is available at https://github.com/Rohit102497/HaphazardInputsAsImages.
URLs: https://github.com/Rohit102497/HaphazardInputsAsImages.
Authors: Chenyu Zhang, Daniil Cherniavskii, Andrii Zadaianchuk, Antonios Tragoudaras, Antonios Vozikis, Thijmen Nijdam, Derck W. E. Prinzhorn, Mark Bodracska, Nicu Sebe, 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: Anurag Kulkarni
Abstract: This paper presents a LiDAR-based object detection system with real-time voice specifications, integrating KITTI's 3D point clouds and RGB images through a multi-modal PointNet framework. It achieves 87.0% validation accuracy on a 3000-sample subset, surpassing a 200-sample baseline of 67.5% by combining spatial and visual data, addressing class imbalance with weighted loss, and refining training via adaptive techniques. A Tkinter prototype provides natural Indian male voice output using Edge TTS (en-IN-PrabhatNeural), alongside 3D visualizations and real-time feedback, enhancing accessibility and safety in autonomous navigation, assistive technology, and beyond. The study offers a detailed methodology, comprehensive experimental analysis, and a broad review of applications and challenges, establishing this work as a scalable advancement in human-computer interaction and environmental perception, aligned with current research trends.
Authors: Xianwei Zhuang, Yuxin Xie, Yufan Deng, Dongchao Yang, Liming Liang, Jinghan Ru, Yuguo Yin, Yuexian Zou
Abstract: In this work, we present VARGPT-v1.1, an advanced unified visual autoregressive model that builds upon our previous framework VARGPT. The model preserves the dual paradigm of next-token prediction for visual understanding and next-scale generation for image synthesis. Specifically, VARGPT-v1.1 integrates: (1) a novel training strategy combining iterative visual instruction tuning with reinforcement learning through Direct Preference Optimization (DPO), (2) an expanded training corpus containing 8.3M visual-generative instruction pairs, (3) an upgraded language model backbone using Qwen2, (4) enhanced image generation resolution, and (5) emergent image editing capabilities without architectural modifications. These advancements enable VARGPT-v1.1 to achieve state-of-the-art performance in multimodal understanding and text-to-image instruction-following tasks, demonstrating significant improvements in both comprehension and generation metrics. Notably, through visual instruction tuning, the model acquires image editing functionality while maintaining architectural consistency with its predecessor, revealing the potential for unified visual understanding, generation, and editing. Our findings suggest that well-designed unified visual autoregressive models can effectively adopt flexible training strategies from large language models (LLMs), exhibiting promising scalability. The codebase and model weights are publicly available at https://github.com/VARGPT-family/VARGPT-v1.1.
Authors: Binh M. Le, Shaoyuan Xu, Jinmiao Fu, Zhishen Huang, Moyan Li, Yanhui Guo, Hongdong Li, Sameera Ramasinghe, Bryan Wang
Abstract: In Visual Document Understanding (VDU) tasks, fine-tuning a pre-trained Vision-Language Model (VLM) with new datasets often falls short in optimizing the vision encoder to identify query-specific regions in text-rich document images. Existing methods that directly inject queries into model layers by modifying the network architecture often struggle to adapt to new datasets with limited annotations. To address this, we introduce QID, a novel, streamlined, architecture-preserving approach that integrates query embeddings into the vision encoder, leading to notable performance gains, particularly in data-scarce fine-tuning scenarios. Specifically, our approach introduces a dual-module framework: a query-aware module that generates a unique query vector to precisely guide the model's focus, as well as a query-agnostic module that captures the positional relationships among tokens, ensuring robust spatial understanding. Notably, both modules operate independently of the vision attention blocks, facilitating targeted learning of query embeddings and enhancing visual semantic identification. Experiments with OCR-free VLMs across multiple datasets demonstrate significant performance improvements using our method, especially in handling text-rich documents in data-scarce environments.
Authors: Jing Gao, Ce Zheng, Laszlo A. Jeni, Zackory Erickson
Abstract: In-bed human mesh recovery can be crucial and enabling for several healthcare applications, including sleep pattern monitoring, rehabilitation support, and pressure ulcer prevention. However, it is difficult to collect large real-world visual datasets in this domain, in part due to privacy and expense constraints, which in turn presents significant challenges for training and deploying deep learning models. Existing in-bed human mesh estimation methods often rely heavily on real-world data, limiting their ability to generalize across different in-bed scenarios, such as varying coverings and environmental settings. To address this, we propose a Sim-to-Real Transfer Framework for in-bed human mesh recovery from overhead depth images, which leverages large-scale synthetic data alongside limited or no real-world samples. We introduce a diffusion model that bridges the gap between synthetic data and real data to support generalization in real-world in-bed pose and body inference scenarios. Extensive experiments and ablation studies validate the effectiveness of our framework, demonstrating significant improvements in robustness and adaptability across diverse healthcare scenarios.
Authors: Shaoyuan Xu, Yang Cheng, Qian Lin, Jan P. Allebach
Abstract: Emotion has an important role in daily life, as it helps people better communicate with and understand each other more efficiently. Facial expressions can be classified into 7 categories: angry, disgust, fear, happy, neutral, sad and surprise. How to detect and recognize these seven emotions has become a popular topic in the past decade. In this paper, we develop an emotion recognition system that can apply emotion recognition on both still images and real-time videos by using deep learning. We build our own emotion recognition classification and regression system from scratch, which includes dataset collection, data preprocessing , model training and testing. Given a certain image or a real-time video, our system is able to show the classification and regression results for all of the 7 emotions. The proposed system is tested on 2 different datasets, and achieved an accuracy of over 80\%. Moreover, the result obtained from real-time testing proves the feasibility of implementing convolutional neural networks in real time to detect emotions accurately and efficiently.
Authors: Junying Wang, Jingyuan Liu, Xin Sun, Krishna Kumar Singh, Zhixin Shu, He Zhang, Jimei Yang, Nanxuan Zhao, Tuanfeng Y. Wang, Simon S. Chen, Ulrich Neumann, Jae Shin Yoon
Abstract: This paper introduces Comprehensive Relighting, the first all-in-one approach that can both control and harmonize the lighting from an image or video of humans with arbitrary body parts from any scene. Building such a generalizable model is extremely challenging due to the lack of dataset, restricting existing image-based relighting models to a specific scenario (e.g., face or static human). To address this challenge, we repurpose a pre-trained diffusion model as a general image prior and jointly model the human relighting and background harmonization in the coarse-to-fine framework. To further enhance the temporal coherence of the relighting, we introduce an unsupervised temporal lighting model that learns the lighting cycle consistency from many real-world videos without any ground truth. In inference time, our temporal lighting module is combined with the diffusion models through the spatio-temporal feature blending algorithms without extra training; and we apply a new guided refinement as a post-processing to preserve the high-frequency details from the input image. In the experiments, Comprehensive Relighting shows a strong generalizability and lighting temporal coherence, outperforming existing image-based human relighting and harmonization methods.
Authors: Shaoyuan Xu, Cheng Lu, Mark Shaw, Peter Bauer, Jan P. Allebach
Abstract: Digital copiers and printers are widely used nowadays. One of the most important things people care about is copying or printing quality. In order to improve it, we previously came up with an SVM-based classification method to classify images with only text, only pictures or a mixture of both based on the fact that modern copiers and printers are equipped with processing pipelines designed specifically for different kinds of images. However, in some other applications, we need to distinguish more than three classes. In this paper, we develop a more advanced SVM-based classification method using four more new features to classify 5 types of images which are text, picture, mixed, receipt and highlight.
Authors: Yiran Xu, Siqi Xie, Zhuofang Li, Harris Shadmany, Yinxiao Li, Luciano Sbaiz, Miaosen Wang, Junjie Ke, Jose Lezama, Hang Qi, Han Zhang, Jesse Berent, Ming-Hsuan Yang, Irfan Essa, Jia-Bin Huang, Feng Yang
Abstract: Image retargeting aims to change the aspect-ratio of an image while maintaining its content and structure with less visual artifacts. Existing methods still generate many artifacts or fail to maintain original content or structure. To address this, we introduce HALO, an end-to-end trainable solution for image retargeting. Since humans are more sensitive to distortions in salient areas than non-salient areas of an image, HALO decomposes the input image into salient/non-salient layers and applies different wrapping fields to different layers. To further minimize the structure distortion in the output images, we propose perceptual structure similarity loss which measures the structure similarity between input and output images and aligns with human perception. Both quantitative results and a user study on the RetargetMe dataset show that HALO achieves SOTA. Especially, our method achieves an 18.4% higher user preference compared to the baselines on average.
Authors: Huiming Sun, Yikang Li, Kangning Yang, Ruineng Li, Daitao Xing, Yangbo Xie, Lan Fu, Kaiyu Zhang, Ming Chen, Jiaming Ding, Jiang Geng, Jie Cai, Zibo Meng, Chiuman Ho
Abstract: Inpainting for real-world human and pedestrian removal in high-resolution video clips presents significant challenges, particularly in achieving high-quality outcomes, ensuring temporal consistency, and managing complex object interactions that involve humans, their belongings, and their shadows. In this paper, we introduce VIP (Video Inpainting Pipeline), a novel promptless video inpainting framework for real-world human removal applications. VIP enhances a state-of-the-art text-to-video model with a motion module and employs a Variational Autoencoder (VAE) for progressive denoising in the latent space. Additionally, we implement an efficient human-and-belongings segmentation for precise mask generation. Sufficient experimental results demonstrate that VIP achieves superior temporal consistency and visual fidelity across diverse real-world scenarios, surpassing state-of-the-art methods on challenging datasets. Our key contributions include the development of the VIP pipeline, a reference frame integration technique, and the Dual-Fusion Latent Segment Refinement method, all of which address the complexities of inpainting in long, high-resolution video sequences.
Authors: Joel Sol, Shadi Alijani, Homayoun Najjaran
Abstract: This paper introduces DRANet-SWD, an extension of existing work that disentangles content and style representations of images for unsupervised domain adaptation (UDA). The approach builds upon DRANet by incorporating the sliced Wasserstein discrepancy (SWD) as a style loss instead of the traditional Gram matrix loss. The potential advantages of SWD over the Gram matrix loss for capturing style variations in domain adaptation are investigated. Experiments using digit classification datasets and driving scenario segmentation validate the method, demonstrating that DRANet-SWD enhances performance. Results indicate that SWD provides a more robust statistical comparison of feature distributions, leading to better style adaptation. These findings highlight the effectiveness of SWD in refining feature alignment and improving domain adaptation tasks across these benchmarks. Our code can be found here.
Authors: Reef Alturki, Adrian Hilton, Jean-Yves Guillemaut
Abstract: In spite of the recent advancements in multi-object tracking, occlusion poses a significant challenge. Multi-camera setups have been used to address this challenge by providing a comprehensive coverage of the scene. Recent multi-view pedestrian detection models have highlighted the potential of an early-fusion strategy, projecting feature maps of all views to a common ground plane or the Bird's Eye View (BEV), and then performing detection. This strategy has been shown to improve both detection and tracking performance. However, the perspective transformation results in significant distortion on the ground plane, affecting the robustness of the appearance features of the pedestrians. To tackle this limitation, we propose a novel model that incorporates attention mechanisms in a multi-view pedestrian tracking scenario. Our model utilizes an early-fusion strategy for detection, and a cross-attention mechanism to establish robust associations between pedestrians in different frames, while efficiently propagating pedestrian features across frames, resulting in a more robust feature representation for each pedestrian. Extensive experiments demonstrate that our model outperforms state-of-the-art models, with an IDF1 score of $96.1\%$ on Wildtrack dataset, and $85.7\%$ on MultiviewX dataset.
Authors: Hyun-Ho Choi, Kangsoo Kim, Ki-Ho Lee, Kisong Lee
Abstract: Accurate and real-time three-dimensional (3D) pose estimation is challenging in resource-constrained and dynamic environments owing to its high computational complexity. To address this issue, this study proposes a novel cooperative inference method for real-time 3D human pose estimation in mobile edge computing (MEC) networks. In the proposed method, multiple end devices equipped with lightweight inference models employ dual confidence thresholds to filter ambiguous images. Only the filtered images are offloaded to an edge server with a more powerful inference model for re-evaluation, thereby improving the estimation accuracy under computational and communication constraints. We numerically analyze the performance of the proposed inference method in terms of the inference accuracy and end-to-end delay and formulate a joint optimization problem to derive the optimal confidence thresholds and transmission time for each device, with the objective of minimizing the mean per-joint position error (MPJPE) while satisfying the required end-to-end delay constraint. To solve this problem, we demonstrate that minimizing the MPJPE is equivalent to maximizing the sum of the inference accuracies for all devices, decompose the problem into manageable subproblems, and present a low-complexity optimization algorithm to obtain a near-optimal solution. The experimental results show that a trade-off exists between the MPJPE and end-to-end delay depending on the confidence thresholds. Furthermore, the results confirm that the proposed cooperative inference method achieves a significant reduction in the MPJPE through the optimal selection of confidence thresholds and transmission times, while consistently satisfying the end-to-end delay requirement in various MEC environments.
Authors: Haishan Wang, Mohammad Hassan Vali, Arno Solin
Abstract: 3D Gaussian Splatting (3DGS) has demonstrated remarkable effectiveness in 3D reconstruction, achieving high-quality results with real-time radiance field rendering. However, a key challenge is the substantial storage cost: reconstructing a single scene typically requires millions of Gaussian splats, each represented by 59 floating-point parameters, resulting in approximately 1~GB of memory. To address this challenge, we propose a compression method by building separate attribute codebooks and storing only discrete code indices. Specifically, we employ noise-substituted vector quantization technique to jointly train the codebooks and model features, ensuring consistency between gradient descent optimization and parameter discretization. Our method reduces the memory consumption efficiently (around $45\times$) while maintaining competitive reconstruction quality on standard 3D benchmark scenes. Experiments on different codebook sizes show the trade-off between compression ratio and image quality. Furthermore, the trained compressed model remains fully compatible with popular 3DGS viewers and enables faster rendering speed, making it well-suited for practical applications.
Authors: Pascal Chang, Jingwei Tang, Markus Gross, Vinicius C. Azevedo
Abstract: Video editing and generation methods often rely on pre-trained image-based diffusion models. During the diffusion process, however, the reliance on rudimentary noise sampling techniques that do not preserve correlations present in subsequent frames of a video is detrimental to the quality of the results. This either produces high-frequency flickering, or texture-sticking artifacts that are not amenable to post-processing. With this in mind, we propose a novel method for preserving temporal correlations in a sequence of noise samples. This approach is materialized by a novel noise representation, dubbed $\int$-noise (integral noise), that reinterprets individual noise samples as a continuously integrated noise field: pixel values do not represent discrete values, but are rather the integral of an underlying infinite-resolution noise over the pixel area. Additionally, we propose a carefully tailored transport method that uses $\int$-noise to accurately advect noise samples over a sequence of frames, maximizing the correlation between different frames while also preserving the noise properties. Our results demonstrate that the proposed $\int$-noise can be used for a variety of tasks, such as video restoration, surrogate rendering, and conditional video generation. See https://warpyournoise.github.io/ for video results.
Authors: Prashant Kumar, Dheeraj Vattikonda, Kshitij Madhav Bhat, Kunal Dargan, Prem Kalra
Abstract: The widespread adoption of learning-based methods for the LiDAR makes autonomous vehicles vulnerable to adversarial attacks through adversarial \textit{point injections (PiJ)}. It poses serious security challenges for navigation and map generation. Despite its critical nature, no major work exists that studies learning-based attacks on LiDAR-based SLAM. Our work proposes SLACK, an end-to-end deep generative adversarial model to attack LiDAR scans with several point injections without deteriorating LiDAR quality. To facilitate SLACK, we design a novel yet simple autoencoder that augments contrastive learning with segmentation-based attention for precise reconstructions. SLACK demonstrates superior performance on the task of \textit{point injections (PiJ)} compared to the best baselines on KITTI and CARLA-64 dataset while maintaining accurate scan quality. We qualitatively and quantitatively demonstrate PiJ attacks using a fraction of LiDAR points. It severely degrades navigation and map quality without deteriorating the LiDAR scan quality.
Authors: Zhen Hao Sia, Yogesh Singh Rawat
Abstract: In this work, we focus on scaling open-vocabulary action detection. Existing approaches for action detection are predominantly limited to closed-set scenarios and rely on complex, parameter-heavy architectures. Extending these models to the open-vocabulary setting poses two key challenges: (1) the lack of large-scale datasets with many action classes for robust training, and (2) parameter-heavy adaptations to a pretrained vision-language contrastive model to convert it for detection, risking overfitting the additional non-pretrained parameters to base action classes. Firstly, we introduce an encoder-only multimodal model for video action detection, reducing the reliance on parameter-heavy additions for video action detection. Secondly, we introduce a simple weakly supervised training strategy to exploit an existing closed-set action detection dataset for pretraining. Finally, we depart from the ill-posed base-to-novel benchmark used by prior works in open-vocabulary action detection and devise a new benchmark to evaluate on existing closed-set action detection datasets without ever using them for training, showing novel results to serve as baselines for future work.
Authors: Xuanyu Liu, Huiyun Yao, Jinggui Gao, Zhongyi Guo, Xue Zhang, Yulin Dong
Abstract: Background:Convolutional Neural Networks(CNN) and Vision Transformers(ViT) are the main techniques used in Medical image segmentation. However, CNN is limited to local contextual information, and ViT's quadratic complexity results in significant computational costs. At the same time, equipping the model to distinguish lesion boundaries with varying degrees of severity is also a challenge encountered in skin lesion segmentation. Purpose:This research aims to optimize the balance between computational costs and long-range dependency modelling and achieve excellent generalization across lesions with different degrees of severity. Methods:we propose a lightweight U-shape network that utilizes Vision Fastformer with Fusion Mechanism (VFFM-UNet). We inherit the advantages of Fastformer's additive attention mechanism, combining element-wise product and matrix product for comprehensive feature extraction and channel reduction to save computational costs. In order to accurately identify the lesion boundaries with varying degrees of severity, we designed Fusion Mechanism including Multi-Granularity Fusion and Channel Fusion, which can process the feature maps in the granularity and channel levels to obtain different contextual information. Results:Comprehensive experiments on the ISIC2017, ISIC2018 and PH2 datasets demonstrate that VFFM-UNet outperforms existing state-of-the-art models regarding parameter numbers, computational complexity and segmentation performance. In short, compared to MISSFormer, our model achieves superior segmentation performance while reducing parameter and computation costs by 101x and 15x, respectively. Conclusions:Both quantitative and qualitative analyses show that VFFM-UNet sets a new benchmark by reaching an ideal balance between parameter numbers, computational complexity, and segmentation performance compared to existing state-of-the-art models.
Authors: Ziteng Wei, Qiang He, Bing Li, Feifei Chen, Yun Yang
Abstract: Vision Transformers (ViTs) excel in computer vision tasks but lack flexibility for edge devices' diverse needs. A vital issue is that ViTs pre-trained to cover a broad range of tasks are \textit{over-qualified} for edge devices that usually demand only part of a ViT's knowledge for specific tasks. Their task-specific accuracy on these edge devices is suboptimal. We discovered that small ViTs that focus on device-specific tasks can improve model accuracy and in the meantime, accelerate model inference. This paper presents NuWa, an approach that derives small ViTs from the base ViT for edge devices with specific task requirements. NuWa can transfer task-specific knowledge extracted from the base ViT into small ViTs that fully leverage constrained resources on edge devices to maximize model accuracy with inference latency assurance. Experiments with three base ViTs on three public datasets demonstrate that compared with state-of-the-art solutions, NuWa improves model accuracy by up to $\text{11.83}\%$ and accelerates model inference by 1.29$\times$ - 2.79$\times$. Code for reproduction is available at https://anonymous.4open.science/r/Task_Specific-3A5E.
Authors: Kahim Wong, Jicheng Zhou, Kemou Li, Yain-Whar Si, Xiaowei Wu, Jiantao Zhou
Abstract: The proliferation of AI-generated content brings significant concerns on the forensic and security issues such as source tracing, copyright protection, etc, highlighting the need for effective watermarking technologies. Font-based text watermarking has emerged as an effective solution to embed information, which could ensure copyright, traceability, and compliance of the generated text content. Existing font watermarking methods usually neglect essential font knowledge, which leads to watermarked fonts of low quality and limited embedding capacity. These methods are also vulnerable to real-world distortions, low-resolution fonts, and inaccurate character segmentation. In this paper, we introduce FontGuard, a novel font watermarking model that harnesses the capabilities of font models and language-guided contrastive learning. Unlike previous methods that focus solely on the pixel-level alteration, FontGuard modifies fonts by altering hidden style features, resulting in better font quality upon watermark embedding. We also leverage the font manifold to increase the embedding capacity of our proposed method by generating substantial font variants closely resembling the original font. Furthermore, in the decoder, we employ an image-text contrastive learning to reconstruct the embedded bits, which can achieve desirable robustness against various real-world transmission distortions. FontGuard outperforms state-of-the-art methods by +5.4%, +7.4%, and +5.8% in decoding accuracy under synthetic, cross-media, and online social network distortions, respectively, while improving the visual quality by 52.7% in terms of LPIPS. Moreover, FontGuard uniquely allows the generation of watermarked fonts for unseen fonts without re-training the network. The code and dataset are available at https://github.com/KAHIMWONG/FontGuard.
Authors: Zahid Hassan Tushar, Adeleke Ademakinwa, Jianwu Wang, Zhibo Zhang, Sanjay Purushotham
Abstract: Accurate cloud property retrieval is vital for understanding cloud behavior and its impact on climate, including applications in weather forecasting, climate modeling, and estimating Earth's radiation balance. The Independent Pixel Approximation (IPA), a widely used physics-based approach, simplifies radiative transfer calculations by assuming each pixel is independent of its neighbors. While computationally efficient, IPA has significant limitations, such as inaccuracies from 3D radiative effects, errors at cloud edges, and ineffectiveness for overlapping or heterogeneous cloud fields. Recent AI/ML-based deep learning models have improved retrieval accuracy by leveraging spatial relationships across pixels. However, these models are often memory-intensive, retrieve only a single cloud property, or struggle with joint property retrievals. To overcome these challenges, we introduce CloudUNet with Attention Module (CAM), a compact UNet-based model that employs attention mechanisms to reduce errors in thick, overlapping cloud regions and a specialized loss function for joint retrieval of Cloud Optical Thickness (COT) and Cloud Effective Radius (CER). Experiments on a Large Eddy Simulation (LES) dataset show that our CAM model outperforms state-of-the-art deep learning methods, reducing mean absolute errors (MAE) by 34% for COT and 42% for CER, and achieving 76% and 86% lower MAE for COT and CER retrievals compared to the IPA method.
Authors: Junkai Zhang, Bin Li, Shoujun Zhou, Yue Du
Abstract: Medical Visual Question Answering (Med-VQA) answers clinical questions using medical images, aiding diagnosis. Designing the MedVQA system holds profound importance in assisting clinical diagnosis and enhancing diagnostic accuracy. Building upon this foundation, Hierarchical Medical VQA extends Medical VQA by organizing medical questions into a hierarchical structure and making level-specific predictions to handle fine-grained distinctions. Recently, many studies have proposed hierarchical MedVQA tasks and established datasets, However, several issues still remain: (1) imperfect hierarchical modeling leads to poor differentiation between question levels causing semantic fragmentation across hierarchies. (2) Excessive reliance on implicit learning in Transformer-based cross-modal self-attention fusion methods, which obscures crucial local semantic correlations in medical scenarios. To address these issues, this study proposes a HiCA-VQA method, including two modules: Hierarchical Prompting for fine-grained medical questions and Hierarchical Answer Decoders. The hierarchical prompting module pre-aligns hierarchical text prompts with image features to guide the model in focusing on specific image regions according to question types, while the hierarchical decoder performs separate predictions for questions at different levels to improve accuracy across granularities. The framework also incorporates a cross-attention fusion module where images serve as queries and text as key-value pairs. Experiments on the Rad-Restruct benchmark demonstrate that the HiCA-VQA framework better outperforms existing state-of-the-art methods in answering hierarchical fine-grained questions. This study provides an effective pathway for hierarchical visual question answering systems, advancing medical image understanding.
Authors: Xin Jin, Simon Niklaus, Zhoutong Zhang, Zhihao Xia, Chunle Guo, Yuting Yang, Jiawen Chen, Chongyi Li
Abstract: Denoising is a crucial step in many video processing pipelines such as in interactive editing, where high quality, speed, and user control are essential. While recent approaches achieve significant improvements in denoising quality by leveraging deep learning, they are prone to unexpected failures due to discrepancies between training data distributions and the wide variety of noise patterns found in real-world videos. These methods also tend to be slow and lack user control. In contrast, traditional denoising methods perform reliably on in-the-wild videos and run relatively quickly on modern hardware. However, they require manually tuning parameters for each input video, which is not only tedious but also requires skill. We bridge the gap between these two paradigms by proposing a differentiable denoising pipeline based on traditional methods. A neural network is then trained to predict the optimal denoising parameters for each specific input, resulting in a robust and efficient approach that also supports user control.
Authors: Xuran Ma, Yexin Liu, Yaofu Liu, Xianfeng Wu, Mingzhe Zheng, Zihao Wang, Ser-Nam Lim, Harry Yang
Abstract: Recent advances in diffusion models have demonstrated remarkable capabilities in video generation. However, the computational intensity remains a significant challenge for practical applications. While feature caching has been proposed to reduce the computational burden of diffusion models, existing methods typically overlook the heterogeneous significance of individual blocks, resulting in suboptimal reuse and degraded output quality. To this end, we address this gap by introducing ProfilingDiT, a novel adaptive caching strategy that explicitly disentangles foreground and background-focused blocks. Through a systematic analysis of attention distributions in diffusion models, we reveal a key observation: 1) Most layers exhibit a consistent preference for either foreground or background regions. 2) Predicted noise shows low inter-step similarity initially, which stabilizes as denoising progresses. This finding inspires us to formulate a selective caching strategy that preserves full computation for dynamic foreground elements while efficiently caching static background features. Our approach substantially reduces computational overhead while preserving visual fidelity. Extensive experiments demonstrate that our framework achieves significant acceleration (e.g., 2.01 times speedup for Wan2.1) while maintaining visual fidelity across comprehensive quality metrics, establishing a viable method for efficient video generation.
Authors: Junshan Hu, Jialiang Mao, Zhikang Liu, Zhongpu Xia, Peng Jia, Xianpeng Lang
Abstract: Conventional Vision-Language Models(VLMs) typically utilize a fixed number of vision tokens, regardless of task complexity. This one-size-fits-all strategy introduces notable inefficiencies: using excessive tokens leads to unnecessary computational overhead in simpler tasks, whereas insufficient tokens compromise fine-grained visual comprehension in more complex contexts. To overcome these limitations, we present TokenFLEX, an innovative and adaptable vision-language framework that encodes images into a variable number of tokens for efficient integration with a Large Language Model (LLM). Our approach is underpinned by two pivotal innovations. Firstly, we present a novel training paradigm that enhances performance across varying numbers of vision tokens by stochastically modulating token counts during training. Secondly, we design a lightweight vision token projector incorporating an adaptive pooling layer and SwiGLU, allowing for flexible downsampling of vision tokens and adaptive selection of features tailored to specific token counts. Comprehensive experiments reveal that TokenFLEX consistently outperforms its fixed-token counterparts, achieving notable performance gains across various token counts enhancements of 1.6%, 1.0%, and 0.4% with 64, 144, and 256 tokens, respectively averaged over eight vision-language benchmarks. These results underscore TokenFLEX's remarkable flexibility while maintaining high-performance vision-language understanding.
Authors: Kexin Tian, Jingrui Mao, Yunlong Zhang, Jiwan Jiang, Yang Zhou, Zhengzhong Tu
Abstract: Recent advancements in Vision-Language Models (VLMs) have demonstrated strong potential for autonomous driving tasks. However, their spatial understanding and reasoning-key capabilities for autonomous driving-still exhibit significant limitations. Notably, none of the existing benchmarks systematically evaluate VLMs' spatial reasoning capabilities in driving scenarios. To fill this gap, we propose NuScenes-SpatialQA, the first large-scale ground-truth-based Question-Answer (QA) benchmark specifically designed to evaluate the spatial understanding and reasoning capabilities of VLMs in autonomous driving. Built upon the NuScenes dataset, the benchmark is constructed through an automated 3D scene graph generation pipeline and a QA generation pipeline. The benchmark systematically evaluates VLMs' performance in both spatial understanding and reasoning across multiple dimensions. Using this benchmark, we conduct extensive experiments on diverse VLMs, including both general and spatial-enhanced models, providing the first comprehensive evaluation of their spatial capabilities in autonomous driving. Surprisingly, the experimental results show that the spatial-enhanced VLM outperforms in qualitative QA but does not demonstrate competitiveness in quantitative QA. In general, VLMs still face considerable challenges in spatial understanding and reasoning.
Authors: Hanbo Bi, Yingchao Feng, Boyuan Tong, Mengyu Wang, Haichen Yu, Yongqiang Mao, Hao Chang, Wenhui Diao, Peijin Wang, Yue Yu, Hanyang Peng, Yehong Zhang, Kun Fu, Xian Sun
Abstract: The rapid advancement of foundation models has revolutionized visual representation learning in a self-supervised manner. However, their application in remote sensing (RS) remains constrained by a fundamental gap: existing models predominantly handle single or limited modalities, overlooking the inherently multi-modal nature of RS observations. Optical, synthetic aperture radar (SAR), and multi-spectral data offer complementary insights that significantly reduce the inherent ambiguity and uncertainty in single-source analysis. To bridge this gap, we introduce RingMoE, a unified multi-modal RS foundation model with 14.7 billion parameters, pre-trained on 400 million multi-modal RS images from nine satellites. RingMoE incorporates three key innovations: (1) A hierarchical Mixture-of-Experts (MoE) architecture comprising modal-specialized, collaborative, and shared experts, effectively modeling intra-modal knowledge while capturing cross-modal dependencies to mitigate conflicts between modal representations; (2) Physics-informed self-supervised learning, explicitly embedding sensor-specific radiometric characteristics into the pre-training objectives; (3) Dynamic expert pruning, enabling adaptive model compression from 14.7B to 1B parameters while maintaining performance, facilitating efficient deployment in Earth observation applications. Evaluated across 23 benchmarks spanning six key RS tasks (i.e., classification, detection, segmentation, tracking, change detection, and depth estimation), RingMoE outperforms existing foundation models and sets new SOTAs, demonstrating remarkable adaptability from single-modal to multi-modal scenarios. Beyond theoretical progress, it has been deployed and trialed in multiple sectors, including emergency response, land management, marine sciences, and urban planning.
Authors: Lucas Choi, Ross Greer
Abstract: In this paper, we address a novel image restoration problem relevant to machine learning dataset curation: the detection and removal of noisy mirrored padding artifacts. While data augmentation techniques like padding are necessary for standardizing image dimensions, they can introduce artifacts that degrade model evaluation when datasets are repurposed across domains. We propose a systematic algorithm to precisely delineate the reflection boundary through a minimum mean squared error approach with thresholding and remove reflective padding. Our method effectively identifies the transition between authentic content and its mirrored counterpart, even in the presence of compression or interpolation noise. We demonstrate our algorithm's efficacy on the SHEL5k dataset, showing significant performance improvements in zero-shot object detection tasks using OWLv2, with average precision increasing from 0.47 to 0.61 for hard hat detection and from 0.68 to 0.73 for person detection. By addressing annotation inconsistencies and distorted objects in padded regions, our approach enhances dataset integrity, enabling more reliable model evaluation across computer vision tasks.
Authors: Shabnam Choudhury, Yash Salunkhe, Sarthak Mehrotra, Biplab Banerjee
Abstract: The rapid expansion of remote sensing image archives demands the development of strong and efficient techniques for content-based image retrieval (RS-CBIR). This paper presents REJEPA (Retrieval with Joint-Embedding Predictive Architecture), an innovative self-supervised framework designed for unimodal RS-CBIR. REJEPA utilises spatially distributed context token encoding to forecast abstract representations of target tokens, effectively capturing high-level semantic features and eliminating unnecessary pixel-level details. In contrast to generative methods that focus on pixel reconstruction or contrastive techniques that depend on negative pairs, REJEPA functions within feature space, achieving a reduction in computational complexity of 40-60% when compared to pixel-reconstruction baselines like Masked Autoencoders (MAE). To guarantee strong and varied representations, REJEPA incorporates Variance-Invariance-Covariance Regularisation (VICReg), which prevents encoder collapse by promoting feature diversity and reducing redundancy. The method demonstrates an estimated enhancement in retrieval accuracy of 5.1% on BEN-14K (S1), 7.4% on BEN-14K (S2), 6.0% on FMoW-RGB, and 10.1% on FMoW-Sentinel compared to prominent SSL techniques, including CSMAE-SESD, Mask-VLM, SatMAE, ScaleMAE, and SatMAE++, on extensive RS benchmarks BEN-14K (multispectral and SAR data), FMoW-RGB and FMoW-Sentinel. Through effective generalisation across sensor modalities, REJEPA establishes itself as a sensor-agnostic benchmark for efficient, scalable, and precise RS-CBIR, addressing challenges like varying resolutions, high object density, and complex backgrounds with computational efficiency.
Authors: Zeyang Zheng, Arman Hosseini, Dong Chen, Omid Shoghli, Arsalan Heydarian
Abstract: The increasing adoption of electric scooters (e-scooters) in urban areas has coincided with a rise in traffic accidents and injuries, largely due to their small wheels, lack of suspension, and sensitivity to uneven surfaces. While deep learning-based object detection has been widely used to improve automobile safety, its application for e-scooter obstacle detection remains unexplored. This study introduces a novel ground obstacle detection system for e-scooters, integrating an RGB camera, and a depth camera to enhance real-time road hazard detection. Additionally, the Inertial Measurement Unit (IMU) measures linear vertical acceleration to identify surface vibrations, guiding the selection of six obstacle categories: tree branches, manhole covers, potholes, pine cones, non-directional cracks, and truncated domes. All sensors, including the RGB camera, depth camera, and IMU, are integrated within the Intel RealSense Camera D435i. A deep learning model powered by YOLO detects road hazards and utilizes depth data to estimate obstacle proximity. Evaluated on the seven hours of naturalistic riding dataset, the system achieves a high mean average precision (mAP) of 0.827 and demonstrates excellent real-time performance. This approach provides an effective solution to enhance e-scooter safety through advanced computer vision and data fusion. The dataset is accessible at https://zenodo.org/records/14583718, and the project code is hosted on https://github.com/Zeyang-Zheng/Real-Time-Roadway-Obstacle-Detection-for-Electric-Scooters.
URLs: https://zenodo.org/records/14583718,, https://github.com/Zeyang-Zheng/Real-Time-Roadway-Obstacle-Detection-for-Electric-Scooters.
Authors: Yuki Kawana, Tatsuya Harada
Abstract: We propose an end-to-end trainable, cross-category method for reconstructing multiple man-made articulated objects from a single RGBD image, focusing on part-level shape reconstruction and pose and kinematics estimation. We depart from previous works that rely on learning instance-level latent space, focusing on man-made articulated objects with predefined part counts. Instead, we propose a novel alternative approach that employs part-level representation, representing instances as combinations of detected parts. While our detect-then-group approach effectively handles instances with diverse part structures and various part counts, it faces issues of false positives, varying part sizes and scales, and an increasing model size due to end-to-end training. To address these challenges, we propose 1) test-time kinematics-aware part fusion to improve detection performance while suppressing false positives, 2) anisotropic scale normalization for part shape learning to accommodate various part sizes and scales, and 3) a balancing strategy for cross-refinement between feature space and output space to improve part detection while maintaining model size. Evaluation on both synthetic and real data demonstrates that our method successfully reconstructs variously structured multiple instances that previous works cannot handle, and outperforms prior works in shape reconstruction and kinematics estimation.
Authors: Shabnam Choudhury, Akhil Vasim, Michael Schmitt, Biplab Banerjee
Abstract: Masked Image Modeling (MIM) is a self-supervised learning technique that involves masking portions of an image, such as pixels, patches, or latent representations, and training models to predict the missing information using the visible context. This approach has emerged as a cornerstone in self-supervised learning, unlocking new possibilities in visual understanding by leveraging unannotated data for pre-training. In remote sensing, MIM addresses challenges such as incomplete data caused by cloud cover, occlusions, and sensor limitations, enabling applications like cloud removal, multi-modal data fusion, and super-resolution. By synthesizing and critically analyzing recent advancements, this survey (MIMRS) is a pioneering effort to chart the landscape of mask image modeling in remote sensing. We highlight state-of-the-art methodologies, applications, and future research directions, providing a foundational review to guide innovation in this rapidly evolving field.
Authors: Sandra Bergmann, Fabian Brand, Christian Riess
Abstract: The JPEG standard was vastly successful. Currently, the first AI-based compression method ``JPEG AI'' will be standardized. JPEG AI brings remarkable benefits. JPEG AI images exhibit impressive image quality at bitrates that are an order of magnitude lower than images compressed with traditional JPEG. However, forensic analysis of JPEG AI has to be completely re-thought: forensic tools for traditional JPEG do not transfer to JPEG AI, and artifacts from JPEG AI are easily confused with artifacts from artificially generated images (``DeepFakes''). This creates a need for novel forensic approaches to detection and distinction of JPEG AI images. In this work, we make a first step towards a forensic JPEG AI toolset. We propose three cues for forensic algorithms for JPEG AI. These algorithms address three forensic questions: first, we show that the JPEG AI preprocessing introduces correlations in the color channels that do not occur in uncompressed images. Second, we show that repeated compression of JPEG AI images leads to diminishing distortion differences. This can be used to detect recompression, in a spirit similar to some classic JPEG forensics methods. Third, we show that the quantization of JPEG AI images in the latent space can be used to distinguish real images with JPEG AI compression from synthetically generated images. The proposed methods are interpretable for a forensic analyst, and we hope that they inspire further research in the forensics of AI-compressed images.
Authors: Xin Zhang, Robby T. Tan
Abstract: Vision Foundation Models (VFMs) and Vision-Language Models (VLMs) have gained traction in Domain Generalized Semantic Segmentation (DGSS) due to their strong generalization capabilities. However, existing DGSS methods often rely exclusively on either VFMs or VLMs, overlooking their complementary strengths. VFMs (e.g., DINOv2) excel at capturing fine-grained features, while VLMs (e.g., CLIP) provide robust text alignment but struggle with coarse granularity. Despite their complementary strengths, effectively integrating VFMs and VLMs with attention mechanisms is challenging, as the increased patch tokens complicate long-sequence modeling. To address this, we propose MFuser, a novel Mamba-based fusion framework that efficiently combines the strengths of VFMs and VLMs while maintaining linear scalability in sequence length. MFuser consists of two key components: MVFuser, which acts as a co-adapter to jointly fine-tune the two models by capturing both sequential and spatial dynamics; and MTEnhancer, a hybrid attention-Mamba module that refines text embeddings by incorporating image priors. Our approach achieves precise feature locality and strong text alignment without incurring significant computational overhead. Extensive experiments demonstrate that MFuser significantly outperforms state-of-the-art DGSS methods, achieving 68.20 mIoU on synthetic-to-real and 71.87 mIoU on real-to-real benchmarks. The code is available at https://github.com/devinxzhang/MFuser.
Authors: Jiaxin Guo, Wenzhen Dong, Tianyu Huang, Hao Ding, Ziyi Wang, Haomin Kuang, Qi Dou, Yun-Hui Liu
Abstract: Reconstructing 3D scenes from monocular surgical videos can enhance surgeon's perception and therefore plays a vital role in various computer-assisted surgery tasks. However, achieving scale-consistent reconstruction remains an open challenge due to inherent issues in endoscopic videos, such as dynamic deformations and textureless surfaces. Despite recent advances, current methods either rely on calibration or instrument priors to estimate scale, or employ SfM-like multi-stage pipelines, leading to error accumulation and requiring offline optimization. In this paper, we present Endo3R, a unified 3D foundation model for online scale-consistent reconstruction from monocular surgical video, without any priors or extra optimization. Our model unifies the tasks by predicting globally aligned pointmaps, scale-consistent video depths, and camera parameters without any offline optimization. The core contribution of our method is expanding the capability of the recent pairwise reconstruction model to long-term incremental dynamic reconstruction by an uncertainty-aware dual memory mechanism. The mechanism maintains history tokens of both short-term dynamics and long-term spatial consistency. Notably, to tackle the highly dynamic nature of surgical scenes, we measure the uncertainty of tokens via Sampson distance and filter out tokens with high uncertainty. Regarding the scarcity of endoscopic datasets with ground-truth depth and camera poses, we further devise a self-supervised mechanism with a novel dynamics-aware flow loss. Abundant experiments on SCARED and Hamlyn datasets demonstrate our superior performance in zero-shot surgical video depth prediction and camera pose estimation with online efficiency. Project page: https://wrld.github.io/Endo3R/.
Authors: Simrandeep Singh, Shreya Bansal, Abdulmotaleb El Saddik, Mukesh Saini
Abstract: The rapid advancement of artificial intelligence (AI) has reshaped the field of natural language processing (NLP), with models like OpenAI ChatGPT and DeepSeek AI. Although ChatGPT established a strong foundation for conversational AI, DeepSeek AI introduces significant improvements in architecture, performance, and ethical considerations. This paper presents a detailed analysis of the evolution from ChatGPT to DeepSeek AI, highlighting their technical differences, practical applications, and broader implications for AI development. To assess their capabilities, we conducted a case study using a predefined set of multiple choice questions in various domains, evaluating the strengths and limitations of each model. By examining these aspects, we provide valuable insight into the future trajectory of AI, its potential to transform industries, and key research directions for improving AI-driven language models.
Authors: Jungpil Shin, Abu Saleh Musa Miah, Sota Konnai, Shu Hoshitaka, Pankoo Kim
Abstract: Hand gesture recognition using multichannel surface electromyography (sEMG) is challenging due to unstable predictions and inefficient time-varying feature enhancement. To overcome the lack of signal based time-varying feature problems, we propose a lightweight squeeze-excitation deep learning-based multi stream spatial temporal dynamics time-varying feature extraction approach to build an effective sEMG-based hand gesture recognition system. Each branch of the proposed model was designed to extract hierarchical features, capturing both global and detailed spatial-temporal relationships to ensure feature effectiveness. The first branch, utilizing a Bidirectional-TCN (Bi-TCN), focuses on capturing long-term temporal dependencies by modelling past and future temporal contexts, providing a holistic view of gesture dynamics. The second branch, incorporating a 1D Convolutional layer, separable CNN, and Squeeze-and-Excitation (SE) block, efficiently extracts spatial-temporal features while emphasizing critical feature channels, enhancing feature relevance. The third branch, combining a Temporal Convolutional Network (TCN) and Bidirectional LSTM (BiLSTM), captures bidirectional temporal relationships and time-varying patterns. Outputs from all branches are fused using concatenation to capture subtle variations in the data and then refined with a channel attention module, selectively focusing on the most informative features while improving computational efficiency. The proposed model was tested on the Ninapro DB2, DB4, and DB5 datasets, achieving accuracy rates of 96.41%, 92.40%, and 93.34%, respectively. These results demonstrate the capability of the system to handle complex sEMG dynamics, offering advancements in prosthetic limb control and human-machine interface technologies with significant implications for assistive technologies.
Authors: Yasmine Mustafa, Mohamed Elmahallawy, Tie Luo
Abstract: Alzheimer's disease (AD) leads to progressive cognitive decline, making early detection crucial for effective intervention. While deep learning models have shown high accuracy in AD diagnosis, their lack of interpretability limits clinical trust and adoption. This paper introduces a novel pre-model approach leveraging Jacobian Maps (JMs) within a multi-modal framework to enhance explainability and trustworthiness in AD detection. By capturing localized brain volume changes, JMs establish meaningful correlations between model predictions and well-known neuroanatomical biomarkers of AD. We validate JMs through experiments comparing a 3D CNN trained on JMs versus on traditional preprocessed data, which demonstrates superior accuracy. We also employ 3D Grad-CAM analysis to provide both visual and quantitative insights, further showcasing improved interpretability and diagnostic reliability.
Authors: Ibne Farabi Shihab, Anuj Sharma
Abstract: Traffic crash detection in long-form surveillance videos is critical for emergency response and infrastructure planning but remains difficult due to the brief and rare nature of crash events. We introduce HybridMamba, a novel architecture that combines visual transformers with state-space temporal modeling to achieve accurate crash time localization. Our method uses multi-level token compression and hierarchical temporal processing to remain computationally efficient without sacrificing temporal resolution. Evaluated on a large-scale dataset from the Iowa Department of Transportation, HybridMamba achieves a mean absolute error of 1.50 seconds, with 65.2 percent of predictions within one second of the ground truth. It outperforms recent video-language models such as TimeChat and VideoLLaMA2 by up to 2.8 seconds, while using significantly fewer parameters. Our results demonstrate strong generalization across videos ranging from 2 to 40 minutes in diverse conditions. HybridMamba offers a robust and efficient solution for fine-grained temporal localization in traffic surveillance. The code will be released upon publication.
Authors: Marius Graumann (German Aerospace Center), Jan Marius St\"urmer (German Aerospace Center), Tobias Koch (German Aerospace Center)
Abstract: Nowadays, a lot of old floor plans exist in printed form or are stored as scanned raster images. Slight rotations or shifts may occur during scanning. Bringing floor plans of this form into a machine readable form to enable further use, still poses a problem. Therefore, we propose an end-to-end pipeline that pre-processes the image and leverages a novel approach to create a region adjacency graph (RAG) from the pre-processed image and predict its nodes. By incorporating normalization steps into the RAG feature extraction, we significantly improved the rotation invariance of the RAG feature calculation. Moreover, applying our method leads to an improved F1 score and IoU on rotated data. Furthermore, we proposed a wall splitting algorithm for partitioning walls into segments associated with the corresponding rooms.
Authors: Piet Br\"ommel, Dominik Br\"amer, Oliver Urbann, Diana Kleingarn
Abstract: The localization of moving robots depends on the availability of good features from the environment. Sensor systems like Lidar are popular, but unique features can also be extracted from images of the ground. This work presents the Keypoint Localization Framework (KOALA), which utilizes deep neural networks that extract sufficient features from an industrial floor for accurate localization without having readable markers. For this purpose, we use a floor covering that can be produced as cheaply as common industrial floors. Although we do not use any filtering, prior, or temporal information, we can estimate our position in 75.7 % of all images with a mean position error of 2 cm and a rotation error of 2.4 %. Thus, the robot kidnapping problem can be solved with high precision in every frame, even while the robot is moving. Furthermore, we show that our framework with our detector and descriptor combination is able to outperform comparable approaches.
Authors: Yimin Wei, Aoran Xiao, Yexian Ren, Yuting Zhu, Hongruixuan Chen, Junshi Xia, Naoto Yokoya
Abstract: Synthetic Aperture Radar (SAR) is a crucial remote sensing technology, enabling all-weather, day-and-night observation with strong surface penetration for precise and continuous environmental monitoring and analysis. However, SAR image interpretation remains challenging due to its complex physical imaging mechanisms and significant visual disparities from human perception. Recently, Vision-Language Models (VLMs) have demonstrated remarkable success in RGB image understanding, offering powerful open-vocabulary interpretation and flexible language interaction. However, their application to SAR images is severely constrained by the absence of SAR-specific knowledge in their training distributions, leading to suboptimal performance. To address this limitation, we introduce SARLANG-1M, a large-scale benchmark tailored for multimodal SAR image understanding, with a primary focus on integrating SAR with textual modality. SARLANG-1M comprises more than 1 million high-quality SAR image-text pairs collected from over 59 cities worldwide. It features hierarchical resolutions (ranging from 0.1 to 25 meters), fine-grained semantic descriptions (including both concise and detailed captions), diverse remote sensing categories (1,696 object types and 16 land cover classes), and multi-task question-answering pairs spanning seven applications and 1,012 question types. Extensive experiments on mainstream VLMs demonstrate that fine-tuning with SARLANG-1M significantly enhances their performance in SAR image interpretation, reaching performance comparable to human experts. The dataset and code will be made publicly available at https://github.com/Jimmyxichen/SARLANG-1M.
Authors: Shuxiao Ding, Yutong Yang, Julian Wiederer, Markus Braun, Peizheng Li, Juergen Gall, Bin Yang
Abstract: Query denoising has become a standard training strategy for DETR-based detectors by addressing the slow convergence issue. Besides that, query denoising can be used to increase the diversity of training samples for modeling complex scenarios which is critical for Multi-Object Tracking (MOT), showing its potential in MOT application. Existing approaches integrate query denoising within the tracking-by-attention paradigm. However, as the denoising process only happens within the single frame, it cannot benefit the tracker to learn temporal-related information. In addition, the attention mask in query denoising prevents information exchange between denoising and object queries, limiting its potential in improving association using self-attention. To address these issues, we propose TQD-Track, which introduces Temporal Query Denoising (TQD) tailored for MOT, enabling denoising queries to carry temporal information and instance-specific feature representation. We introduce diverse noise types onto denoising queries that simulate real-world challenges in MOT. We analyze our proposed TQD for different tracking paradigms, and find out the paradigm with explicit learned data association module, e.g. tracking-by-detection or alternating detection and association, benefit from TQD by a larger margin. For these paradigms, we further design an association mask in the association module to ensure the consistent interaction between track and detection queries as during inference. Extensive experiments on the nuScenes dataset demonstrate that our approach consistently enhances different tracking methods by only changing the training process, especially the paradigms with explicit association module.
Authors: Gia-Nghia Tran, Quang-Huy Che, Trong-Tai Dam Vu, Bich-Nga Pham, Vinh-Tiep Nguyen, Trung-Nghia Le, Minh-Triet Tran
Abstract: Generating multiple new concepts remains a challenging problem in the text-to-image task. Current methods often overfit when trained on a small number of samples and struggle with attribute leakage, particularly for class-similar subjects (e.g., two specific dogs). In this paper, we introduce Fuse-and-Refine (FaR), a novel approach that tackles these challenges through two key contributions: Concept Fusion technique and Localized Refinement loss function. Concept Fusion systematically augments the training data by separating reference subjects from backgrounds and recombining them into composite images to increase diversity. This augmentation technique tackles the overfitting problem by mitigating the narrow distribution of the limited training samples. In addition, Localized Refinement loss function is introduced to preserve subject representative attributes by aligning each concept's attention map to its correct region. This approach effectively prevents attribute leakage by ensuring that the diffusion model distinguishes similar subjects without mixing their attention maps during the denoising process. By fine-tuning specific modules at the same time, FaR balances the learning of new concepts with the retention of previously learned knowledge. Empirical results show that FaR not only prevents overfitting and attribute leakage while maintaining photorealism, but also outperforms other state-of-the-art methods.
Authors: Mathis Kruse, Bodo Rosenhahn
Abstract: With more well-performing anomaly detection methods proposed, many of the single-view tasks have been solved to a relatively good degree. However, real-world production scenarios often involve complex industrial products, whose properties may not be fully captured by one single image. While normalizing flow based approaches already work well in single-camera scenarios, they currently do not make use of the priors in multi-view data. We aim to bridge this gap by using these flow-based models as a strong foundation and propose Multi-Flow, a novel multi-view anomaly detection method. Multi-Flow makes use of a novel multi-view architecture, whose exact likelihood estimation is enhanced by fusing information across different views. For this, we propose a new cross-view message-passing scheme, letting information flow between neighboring views. We empirically validate it on the real-world multi-view data set Real-IAD and reach a new state-of-the-art, surpassing current baselines in both image-wise and sample-wise anomaly detection tasks.
Authors: Bram de Wilde, Max T. Rietberg, Guillaume Lajoinie, Jelmer M. Wolterink
Abstract: Generative modeling of anatomical structures plays a crucial role in virtual imaging trials, which allow researchers to perform studies without the costs and constraints inherent to in vivo and phantom studies. For clinical relevance, generative models should allow targeted control to simulate specific patient populations rather than relying on purely random sampling. In this work, we propose a steerable generative model based on implicit neural representations. Implicit neural representations naturally support topology changes, making them well-suited for anatomical structures with varying topology, such as the thyroid. Our model learns a disentangled latent representation, enabling fine-grained control over shape variations. Evaluation includes reconstruction accuracy and anatomical plausibility. Our results demonstrate that the proposed model achieves high-quality shape generation while enabling targeted anatomical modifications.
Authors: Quanxing Xu, Ling Zhou, Xian Zhong, Feifei Zhang, Rubing Huang, Chia-Wen Lin
Abstract: Existing debiasing approaches in Visual Question Answering (VQA) primarily focus on enhancing visual learning, integrating auxiliary models, or employing data augmentation strategies. However, these methods exhibit two major drawbacks. First, current debiasing techniques fail to capture the superior relation between images and texts because prevalent learning frameworks do not enable models to extract deeper correlations from highly contrasting samples. Second, they do not assess the relevance between the input question and image during inference, as no prior work has examined the degree of input relevance in debiasing studies. Motivated by these limitations, we propose a novel framework, Optimized Question-Image Relation Learning (QIRL), which employs a generation-based self-supervised learning strategy. Specifically, two modules are introduced to address the aforementioned issues. The Negative Image Generation (NIG) module automatically produces highly irrelevant question-image pairs during training to enhance correlation learning, while the Irrelevant Sample Identification (ISI) module improves model robustness by detecting and filtering irrelevant inputs, thereby reducing prediction errors. Furthermore, to validate our concept of reducing output errors through filtering unrelated question-image inputs, we propose a specialized metric to evaluate the performance of the ISI module. Notably, our approach is model-agnostic and can be integrated with various VQA models. Extensive experiments on VQA-CPv2 and VQA-v2 demonstrate the effectiveness and generalization ability of our method. Among data augmentation strategies, our approach achieves state-of-the-art results.
Authors: Guide Yang, Chao Hou, Weilong Peng, Xiang Fang, Yongwei Nie, Peican Zhu, Keke Tang
Abstract: Deep neural networks (DNNs) often exhibit overconfidence when encountering out-of-distribution (OOD) samples, posing significant challenges for deployment. Since DNNs are trained on in-distribution (ID) datasets, the information flow of ID samples through DNNs inevitably differs from that of OOD samples. In this paper, we propose an Entropy-based Out-Of-distribution Detection (EOOD) framework. EOOD first identifies specific block where the information flow differences between ID and OOD samples are more pronounced, using both ID and pseudo-OOD samples. It then calculates the conditional entropy on the selected block as the OOD confidence score. Comprehensive experiments conducted across various ID and OOD settings demonstrate the effectiveness of EOOD in OOD detection and its superiority over state-of-the-art methods.
Authors: Denis Coquenet
Abstract: Recent advances in text recognition led to a paradigm shift for page-level recognition, from multi-step segmentation-based approaches to end-to-end attention-based ones. However, the na\"ive character-level autoregressive decoding process results in long prediction times: it requires several seconds to process a single page image on a modern GPU. We propose the Meta Document Attention Network (Meta-DAN) as a novel decoding strategy to reduce the prediction time while enabling a better context modeling. It relies on two main components: windowed queries, to process several transformer queries altogether, enlarging the context modeling with near future; and multi-token predictions, whose goal is to predict several tokens per query instead of only the next one. We evaluate the proposed approach on 10 full-page handwritten datasets and demonstrate state-of-the-art results on average in terms of character error rate. Source code and weights of trained models are available at https://github.com/FactoDeepLearning/meta_dan.
Authors: Edern Le Bot, R\'emi Giraud, Boris Mansencal, Thomas Tourdias, Jos\`e V. Manjon, Pierrick Coup\'e
Abstract: This paper introduces a novel method for brain segmentation using only FLAIR MRIs, specifically targeting cases where access to other imaging modalities is limited. By leveraging existing automatic segmentation methods, we train a network to approximate segmentations, typically obtained from T1-weighted MRIs. Our method, called FLAIRBrainSeg, produces segmentations of 132 structures and is robust to multiple sclerosis lesions. Experiments on both in-domain and out-of-domain datasets demonstrate that our method outperforms modality-agnostic approaches based on image synthesis, the only currently available alternative for performing brain parcellation using FLAIR MRI alone. This technique holds promise for scenarios where T1-weighted MRIs are unavailable and offers a valuable alternative for clinicians and researchers in need of reliable anatomical segmentation.
Authors: Sheng Yang, Tong Zhan, Shichen Qiao, Jicheng Gong, Qing Yang, Yanfeng Lu, Jian Wang
Abstract: Reliable 3D object perception is essential in autonomous driving. Owing to its sensing capabilities in all weather conditions, 4D radar has recently received much attention. However, compared to LiDAR, 4D radar provides much sparser point cloud. In this paper, we propose a 3D object detection method, termed ZFusion, which fuses 4D radar and vision modality. As the core of ZFusion, our proposed FP-DDCA (Feature Pyramid-Double Deformable Cross Attention) fuser complements the (sparse) radar information and (dense) vision information, effectively. Specifically, with a feature-pyramid structure, the FP-DDCA fuser packs Transformer blocks to interactively fuse multi-modal features at different scales, thus enhancing perception accuracy. In addition, we utilize the Depth-Context-Split view transformation module due to the physical properties of 4D radar. Considering that 4D radar has a much lower cost than LiDAR, ZFusion is an attractive alternative to LiDAR-based methods. In typical traffic scenarios like the VoD (View-of-Delft) dataset, experiments show that with reasonable inference speed, ZFusion achieved the state-of-the-art mAP (mean average precision) in the region of interest, while having competitive mAP in the entire area compared to the baseline methods, which demonstrates performance close to LiDAR and greatly outperforms those camera-only methods.
Authors: Mirko Borszukovszki, Ivo Pascal de Jong, Matias Valdenegro-Toro
Abstract: To leverage the full potential of Large Language Models (LLMs) it is crucial to have some information on their answers' uncertainty. This means that the model has to be able to quantify how certain it is in the correctness of a given response. Bad uncertainty estimates can lead to overconfident wrong answers undermining trust in these models. Quite a lot of research has been done on language models that work with text inputs and provide text outputs. Still, since the visual capabilities have been added to these models recently, there has not been much progress on the uncertainty of Visual Language Models (VLMs). We tested three state-of-the-art VLMs on corrupted image data. We found that the severity of the corruption negatively impacted the models' ability to estimate their uncertainty and the models also showed overconfidence in most of the experiments.
Authors: Nasar Iqbal, Niki Martinel
Abstract: Recent advances in convolutional neural networks (CNNs) and transformer-based methods have improved anomaly detection and localization, but challenges persist in precisely localizing small anomalies. While CNNs face limitations in capturing long-range dependencies, transformer architectures often suffer from substantial computational overheads. We introduce a state space model (SSM)-based Pyramidal Scanning Strategy (PSS) for multi-class anomaly detection and localization--a novel approach designed to address the challenge of small anomaly localization. Our method captures fine-grained details at multiple scales by integrating the PSS with a pre-trained encoder for multi-scale feature extraction and a feature-level synthetic anomaly generator. An improvement of $+1\%$ AP for multi-class anomaly localization and a +$1\%$ increase in AU-PRO on MVTec benchmark demonstrate our method's superiority in precise anomaly localization across diverse industrial scenarios. The code is available at https://github.com/iqbalmlpuniud/Pyramid Mamba.
Authors: Antoine Dumoulin, Adnane Boukhayma, Laurence Boissieux, Bharath Bhushan Damodaran, Pierre Hellier, Stefanie Wuhrer
Abstract: Adjusting and deforming 3D garments to body shapes, body motion, and cloth material is an important problem in virtual and augmented reality. Applications are numerous, ranging from virtual change rooms to the entertainment and gaming industry. This problem is challenging as garment dynamics influence geometric details such as wrinkling patterns, which depend on physical input including the wearer's body shape and motion, as well as cloth material features. Existing work studies learning-based modeling techniques to generate garment deformations from example data, and physics-inspired simulators to generate realistic garment dynamics. We propose here a learning-based approach trained on data generated with a physics-based simulator. Compared to prior work, our 3D generative model learns garment deformations for loose cloth geometry, especially for large deformations and dynamic wrinkles driven by body motion and cloth material. Furthermore, the model can be efficiently fitted to observations captured using vision sensors. We propose to leverage the capability of diffusion models to learn fine-scale detail: we model the 3D garment in a 2D parameter space, and learn a latent diffusion model using this representation independent from the mesh resolution. This allows to condition global and local geometric information with body and material information. We quantitatively and qualitatively evaluate our method on both simulated data and data captured with a multi-view acquisition platform. Compared to strong baselines, our method is more accurate in terms of Chamfer distance.
Authors: Xi Wang, Ziqi He, Yang Zhou
Abstract: Traditional diffusion models typically employ a U-Net architecture. Previous studies have unveiled the roles of attention blocks in the U-Net. However, they overlook the dynamic evolution of their importance during the inference process, which hinders their further exploitation to improve image applications. In this study, we first theoretically proved that, re-weighting the outputs of the Transformer blocks within the U-Net is a "free lunch" for improving the signal-to-noise ratio during the sampling process. Next, we proposed Importance Probe to uncover and quantify the dynamic shifts in importance of the Transformer blocks throughout the denoising process. Finally, we design an adaptive importance-based re-weighting schedule tailored to specific image generation and editing tasks. Experimental results demonstrate that, our approach significantly improves the efficiency of the inference process, and enhances the aesthetic quality of the samples with identity consistency. Our method can be seamlessly integrated into any U-Net-based architecture. Code: https://github.com/Hytidel/UNetReweighting
Authors: Seyedeh Sahar Taheri Otaghsara, Reza Rahmanzadeh
Abstract: This study addresses the essential task of medical image segmentation, which involves the automatic identification and delineation of anatomical structures and pathological regions in medical images. Accurate segmentation is crucial in radiology, as it aids in the precise localization of abnormalities such as tumors, thereby enabling effective diagnosis, treatment planning, and monitoring of disease progression. Specifically, the size, shape, and location of tumors can significantly influence clinical decision-making and therapeutic strategies, making accurate segmentation a key component of radiological workflows. However, challenges posed by variations in MRI modalities, image artifacts, and the scarcity of labeled data complicate the segmentation task and impact the performance of traditional models. To overcome these limitations, we propose a novel self-supervised learning Multi-encoder nnU-Net architecture designed to process multiple MRI modalities independently through separate encoders. This approach allows the model to capture modality-specific features before fusing them for the final segmentation, thus improving accuracy. Our Multi-encoder nnU-Net demonstrates exceptional performance, achieving a Dice Similarity Coefficient (DSC) of 93.72%, which surpasses that of other models such as vanilla nnU-Net, SegResNet, and Swin UNETR. By leveraging the unique information provided by each modality, the model enhances segmentation tasks, particularly in scenarios with limited annotated data. Evaluations highlight the effectiveness of this architecture in improving tumor segmentation outcomes.
Authors: Sheng Lian, Dengfeng Pan, Jianlong Cai, Guang-Yong Chen, Zhun Zhong, Zhiming Luo, Shen Zhao, Shuo Li
Abstract: Accurate lumbar spine segmentation is crucial for diagnosing spinal disorders. Existing methods typically use coarse-grained segmentation strategies that lack the fine detail needed for precise diagnosis. Additionally, their reliance on visual-only models hinders the capture of anatomical semantics, leading to misclassified categories and poor segmentation details. To address these limitations, we present ATM-Net, an innovative framework that employs an anatomy-aware, text-guided, multi-modal fusion mechanism for fine-grained segmentation of lumbar substructures, i.e., vertebrae (VBs), intervertebral discs (IDs), and spinal canal (SC). ATM-Net adopts the Anatomy-aware Text Prompt Generator (ATPG) to adaptively convert image annotations into anatomy-aware prompts in different views. These insights are further integrated with image features via the Holistic Anatomy-aware Semantic Fusion (HASF) module, building a comprehensive anatomical context. The Channel-wise Contrastive Anatomy-Aware Enhancement (CCAE) module further enhances class discrimination and refines segmentation through class-wise channel-level multi-modal contrastive learning. Extensive experiments on the MRSpineSeg and SPIDER datasets demonstrate that ATM-Net significantly outperforms state-of-the-art methods, with consistent improvements regarding class discrimination and segmentation details. For example, ATM-Net achieves Dice of 79.39% and HD95 of 9.91 pixels on SPIDER, outperforming the competitive SpineParseNet by 8.31% and 4.14 pixels, respectively.
Authors: Zihao He, Shengchuan Zhang, Runze Hu, Yunhang Shen, Yan Zhang
Abstract: Super-resolution (SR) techniques are critical for enhancing image quality, particularly in scenarios where high-resolution imagery is essential yet limited by hardware constraints. Existing diffusion models for SR have relied predominantly on Gaussian models for noise generation, which often fall short when dealing with the complex and variable texture inherent in natural scenes. To address these deficiencies, we introduce the Bayesian Uncertainty Guided Diffusion Probabilistic Model (BUFF). BUFF distinguishes itself by incorporating a Bayesian network to generate high-resolution uncertainty masks. These masks guide the diffusion process, allowing for the adjustment of noise intensity in a manner that is both context-aware and adaptive. This novel approach not only enhances the fidelity of super-resolved images to their original high-resolution counterparts but also significantly mitigates artifacts and blurring in areas characterized by complex textures and fine details. The model demonstrates exceptional robustness against complex noise patterns and showcases superior adaptability in handling textures and edges within images. Empirical evidence, supported by visual results, illustrates the model's robustness, especially in challenging scenarios, and its effectiveness in addressing common SR issues such as blurring. Experimental evaluations conducted on the DIV2K dataset reveal that BUFF achieves a notable improvement, with a +0.61 increase compared to baseline in SSIM on BSD100, surpassing traditional diffusion approaches by an average additional +0.20dB PSNR gain. These findings underscore the potential of Bayesian methods in enhancing diffusion processes for SR, paving the way for future advancements in the field.
Authors: Ilan Naiman, Emanuel Ben-Baruch, Oron Anschel, Alon Shoshan, Igor Kviatkovsky, Manoj Aggarwal, Gerard Medioni
Abstract: In this work, we introduce long-video masked-embedding autoencoders (LV-MAE), a self-supervised learning framework for long video representation. Our approach treats short- and long-span dependencies as two separate tasks. Such decoupling allows for a more intuitive video processing where short-span spatiotemporal primitives are first encoded and are then used to capture long-range dependencies across consecutive video segments. To achieve this, we leverage advanced off-the-shelf multimodal encoders to extract representations from short segments within the long video, followed by pre-training a masked-embedding autoencoder capturing high-level interactions across segments. LV-MAE is highly efficient to train and enables the processing of much longer videos by alleviating the constraint on the number of input frames. Furthermore, unlike existing methods that typically pre-train on short-video datasets, our approach offers self-supervised pre-training using long video samples (e.g., 20+ minutes video clips) at scale. Using LV-MAE representations, we achieve state-of-the-art results on three long-video benchmarks -- LVU, COIN, and Breakfast -- employing only a simple classification head for either attentive or linear probing. Finally, to assess LV-MAE pre-training and visualize its reconstruction quality, we leverage the video-language aligned space of short video representations to monitor LV-MAE through video-text retrieval.
Authors: Tan Shu, Li Shen
Abstract: Cropland non-agriculturalization refers to the conversion of arable land into non-agricultural uses such as forests, residential areas, and construction sites. This phenomenon not only directly leads to the loss of cropland resources but also poses systemic threats to food security and agricultural sustainability. Accurate identification of cropland and non-cropland areas is crucial for detecting and addressing this issue. Traditional CNNs employ static convolution layers, while dynamic convolution studies demonstrate that adaptively weighting multiple convolutional kernels through attention mechanisms can enhance accuracy. However, existing dynamic convolution methods relying on Global Average Pooling (GAP) for attention weight allocation suffer from information loss, limiting segmentation precision. This paper proposes Frequency-Aware Dynamic Convolution (FADConv) and a Frequency Attention (FAT) module to address these limitations. Building upon the foundational structure of dynamic convolution, we designed FADConv by integrating 2D Discrete Cosine Transform (2D DCT) to capture frequency domain features and fuse them. FAT module generates high-quality attention weights that replace the traditional GAP method,making the combination between dynamic convolution kernels more reasonable.Experiments on the GID and Hi-CNA datasets demonstrate that FADConv significantly improves segmentation accuracy with minimal computational overhead. For instance, ResNet18 with FADConv achieves 1.9% and 2.7% increases in F1-score and IoU for cropland segmentation on GID, with only 58.87M additional MAdds. Compared to other dynamic convolution approaches, FADConv exhibits superior performance in cropland segmentation tasks.
Authors: Gianluca Monaci, Rafael S. Rezende, Romain Deffayet, Gabriela Csurka, Guillaume Bono, Herv\'e D\'ejean, St\'ephane Clinchant, Christian Wolf
Abstract: Methods for navigation based on large-scale learning typically treat each episode as a new problem, where the agent is spawned with a clean memory in an unknown environment. While these generalization capabilities to an unknown environment are extremely important, we claim that, in a realistic setting, an agent should have the capacity of exploiting information collected during earlier robot operations. We address this by introducing a new retrieval-augmented agent, trained with RL, capable of querying a database collected from previous episodes in the same environment and learning how to integrate this additional context information. We introduce a unique agent architecture for the general navigation task, evaluated on ObjectNav, ImageNav and Instance-ImageNav. Our retrieval and context encoding methods are data-driven and heavily employ vision foundation models (FM) for both semantic and geometric understanding. We propose new benchmarks for these settings and we show that retrieval allows zero-shot transfer across tasks and environments while significantly improving performance.
Authors: Boyuan Wang, Runqi Ouyang, Xiaofeng Wang, Zheng Zhu, Guosheng Zhao, Chaojun Ni, Guan Huang, Lihong Liu, Xingang Wang
Abstract: Single-image human reconstruction is vital for digital human modeling applications but remains an extremely challenging task. Current approaches rely on generative models to synthesize multi-view images for subsequent 3D reconstruction and animation. However, directly generating multiple views from a single human image suffers from geometric inconsistencies, resulting in issues like fragmented or blurred limbs in the reconstructed models. To tackle these limitations, we introduce \textbf{HumanDreamer-X}, a novel framework that integrates multi-view human generation and reconstruction into a unified pipeline, which significantly enhances the geometric consistency and visual fidelity of the reconstructed 3D models. In this framework, 3D Gaussian Splatting serves as an explicit 3D representation to provide initial geometry and appearance priority. Building upon this foundation, \textbf{HumanFixer} is trained to restore 3DGS renderings, which guarantee photorealistic results. Furthermore, we delve into the inherent challenges associated with attention mechanisms in multi-view human generation, and propose an attention modulation strategy that effectively enhances geometric details identity consistency across multi-view. Experimental results demonstrate that our approach markedly improves generation and reconstruction PSNR quality metrics by 16.45% and 12.65%, respectively, achieving a PSNR of up to 25.62 dB, while also showing generalization capabilities on in-the-wild data and applicability to various human reconstruction backbone models.
Authors: Kaidong Li, Tianxiao Zhang, Kuan-Chuan Peng, Guanghui Wang
Abstract: 3D object detection is crucial for autonomous driving, leveraging both LiDAR point clouds for precise depth information and camera images for rich semantic information. Therefore, the multi-modal methods that combine both modalities offer more robust detection results. However, efficiently fusing LiDAR points and images remains challenging due to the domain gaps. In addition, the performance of many models is limited by the amount of high quality labeled data, which is expensive to create. The recent advances in foundation models, which use large-scale pre-training on different modalities, enable better multi-modal fusion. Combining the prompt engineering techniques for efficient training, we propose the Prompted Foundational 3D Detector (PF3Det), which integrates foundation model encoders and soft prompts to enhance LiDAR-camera feature fusion. PF3Det achieves the state-of-the-art results under limited training data, improving NDS by 1.19% and mAP by 2.42% on the nuScenes dataset, demonstrating its efficiency in 3D detection.
Authors: Niu Lian, Jun Li, Jinpeng Wang, Ruisheng Luo, Yaowei Wang, Shu-Tao Xia, Bin Chen
Abstract: Self-Supervised Video Hashing (SSVH) compresses videos into hash codes for efficient indexing and retrieval using unlabeled training videos. Existing approaches rely on random frame sampling to learn video features and treat all frames equally. This results in suboptimal hash codes, as it ignores frame-specific information density and reconstruction difficulty. To address this limitation, we propose a new framework, termed AutoSSVH, that employs adversarial frame sampling with hash-based contrastive learning. Our adversarial sampling strategy automatically identifies and selects challenging frames with richer information for reconstruction, enhancing encoding capability. Additionally, we introduce a hash component voting strategy and a point-to-set (P2Set) hash-based contrastive objective, which help capture complex inter-video semantic relationships in the Hamming space and improve the discriminability of learned hash codes. Extensive experiments demonstrate that AutoSSVH achieves superior retrieval efficacy and efficiency compared to state-of-the-art approaches. Code is available at https://github.com/EliSpectre/CVPR25-AutoSSVH.
Authors: Kai Lascheit, Daniel Barath, Marc Pollefeys, Leonidas Guibas, Francis Engelmann
Abstract: Registering human meshes to 3D point clouds is essential for applications such as augmented reality and human-robot interaction but often yields imprecise results due to noise and background clutter in real-world data. We introduce a hybrid approach that incorporates body-part segmentation into the mesh fitting process, enhancing both human pose estimation and segmentation accuracy. Our method first assigns body part labels to individual points, which then guide a two-step SMPL-X fitting: initial pose and orientation estimation using body part centroids, followed by global refinement of the point cloud alignment. Additionally, we demonstrate that the fitted human mesh can refine body part labels, leading to improved segmentation. Evaluations on the cluttered and noisy real-world datasets InterCap, EgoBody, and BEHAVE show that our approach significantly outperforms prior methods in both pose estimation and segmentation accuracy. Code and results are available on our project website: https://segfit.github.io
URLs: https://segfit.github.io
Authors: Yuyang Hu, Suhas Lohit, Ulugbek S. Kamilov, Tim K. Marks
Abstract: Deep learning has achieved some success in addressing the challenge of cloud removal in optical satellite images, by fusing with synthetic aperture radar (SAR) images. Recently, diffusion models have emerged as powerful tools for cloud removal, delivering higher-quality estimation by sampling from cloud-free distributions, compared to earlier methods. However, diffusion models initiate sampling from pure Gaussian noise, which complicates the sampling trajectory and results in suboptimal performance. Also, current methods fall short in effectively fusing SAR and optical data. To address these limitations, we propose Diffusion Bridges for Cloud Removal, DB-CR, which directly bridges between the cloudy and cloud-free image distributions. In addition, we propose a novel multimodal diffusion bridge architecture with a two-branch backbone for multimodal image restoration, incorporating an efficient backbone and dedicated cross-modality fusion blocks to effectively extract and fuse features from synthetic aperture radar (SAR) and optical images. By formulating cloud removal as a diffusion-bridge problem and leveraging this tailored architecture, DB-CR achieves high-fidelity results while being computationally efficient. We evaluated DB-CR on the SEN12MS-CR cloud-removal dataset, demonstrating that it achieves state-of-the-art results.
Authors: Aref Azizpour, Tai D. Nguyen, Matthew C. Stamm
Abstract: Rapid advances in generative AI have enabled the creation of highly realistic synthetic images, which, while beneficial in many domains, also pose serious risks in terms of disinformation, fraud, and other malicious applications. Current synthetic image identification systems are typically static, relying on feature representations learned from known generators; as new generative models emerge, these systems suffer from severe performance degradation. In this paper, we introduce the concept of an autonomous self-adaptive synthetic media identification system -- one that not only detects synthetic images and attributes them to known sources but also autonomously identifies and incorporates novel generators without human intervention. Our approach leverages an open-set identification strategy with an evolvable embedding space that distinguishes between known and unknown sources. By employing an unsupervised clustering method to aggregate unknown samples into high-confidence clusters and continuously refining its decision boundaries, our system maintains robust detection and attribution performance even as the generative landscape evolves. Extensive experiments demonstrate that our method significantly outperforms existing approaches, marking a crucial step toward universal, adaptable forensic systems in the era of rapidly advancing generative models.
Authors: Laziz Hamdi, Amine Tamasna, Pascal Boisson, Thierry Paquet
Abstract: We introduce \textbf{VISTA-OCR} (Vision and Spatially-aware Text Analysis OCR), a lightweight architecture that unifies text detection and recognition within a single generative model. Unlike conventional methods that require separate branches with dedicated parameters for text recognition and detection, our approach leverages a Transformer decoder to sequentially generate text transcriptions and their spatial coordinates in a unified branch. Built on an encoder-decoder architecture, VISTA-OCR is progressively trained, starting with the visual feature extraction phase, followed by multitask learning with multimodal token generation. To address the increasing demand for versatile OCR systems capable of advanced tasks, such as content-based text localization \ref{content_based_localization}, we introduce new prompt-controllable OCR tasks during pre-training.To enhance the model's capabilities, we built a new dataset composed of real-world examples enriched with bounding box annotations and synthetic samples. Although recent Vision Large Language Models (VLLMs) can efficiently perform these tasks, their high computational cost remains a barrier for practical deployment. In contrast, our VISTA$_{\text{omni}}$ variant processes both handwritten and printed documents with only 150M parameters, interactively, by prompting. Extensive experiments on multiple datasets demonstrate that VISTA-OCR achieves better performance compared to state-of-the-art specialized models on standard OCR tasks while showing strong potential for more sophisticated OCR applications, addressing the growing need for interactive OCR systems. All code and annotations for VISTA-OCR will be made publicly available upon acceptance.
Authors: Ciaran Bench, Spencer A. Thomas
Abstract: The quality of synthetically generated images (e.g. those produced by diffusion models) are often evaluated using information about image contents encoded by pretrained auxiliary models. For example, the Fr\'{e}chet Inception Distance (FID) uses embeddings from an InceptionV3 model pretrained to classify ImageNet. The effectiveness of this feature embedding model has considerable impact on the trustworthiness of the calculated metric (affecting its suitability in several domains, including medical imaging). Here, uncertainty quantification (UQ) is used to provide a heuristic measure of the trustworthiness of the feature embedding model and an FID-like metric called the Fr\'{e}chet Autoencoder Distance (FAED). We apply Monte Carlo dropout to a feature embedding model (convolutional autoencoder) to model the uncertainty in its embeddings. The distribution of embeddings for each input are then used to compute a distribution of FAED values. We express uncertainty as the predictive variance of the embeddings as well as the standard deviation of the computed FAED values. We find that their magnitude correlates with the extent to which the inputs are out-of-distribution to the model's training data, providing some validation of its ability to assess the trustworthiness of the FAED.
Authors: Federica Arrigoni, Kathl\'en Kohn, Andrea Fusiello, Tomas Pajdla
Abstract: The concept of viewing graph solvability has gained significant interest in the context of structure-from-motion. A viewing graph is a mathematical structure where nodes are associated to cameras and edges represent the epipolar geometry connecting overlapping views. Solvability studies under which conditions the cameras are uniquely determined by the graph. In this paper we propose a novel framework for analyzing solvability problems based on Algebraic Geometry, demonstrating its potential in understanding structure-from-motion graphs and proving a conjecture that was previously proposed.
Authors: Ting-Hsuan Liao, Yi Zhou, Yu Shen, Chun-Hao Paul Huang, Saayan Mitra, Jia-Bin Huang, Uttaran Bhattacharya
Abstract: We explore how body shapes influence human motion synthesis, an aspect often overlooked in existing text-to-motion generation methods due to the ease of learning a homogenized, canonical body shape. However, this homogenization can distort the natural correlations between different body shapes and their motion dynamics. Our method addresses this gap by generating body-shape-aware human motions from natural language prompts. We utilize a finite scalar quantization-based variational autoencoder (FSQ-VAE) to quantize motion into discrete tokens and then leverage continuous body shape information to de-quantize these tokens back into continuous, detailed motion. Additionally, we harness the capabilities of a pretrained language model to predict both continuous shape parameters and motion tokens, facilitating the synthesis of text-aligned motions and decoding them into shape-aware motions. We evaluate our method quantitatively and qualitatively, and also conduct a comprehensive perceptual study to demonstrate its efficacy in generating shape-aware motions.
Authors: Wulin Xie, Yi-Fan Zhang, Chaoyou Fu, Yang Shi, Bingyan Nie, Hongkai Chen, Zhang Zhang, Liang Wang, Tieniu Tan
Abstract: Existing MLLM benchmarks face significant challenges in evaluating Unified MLLMs (U-MLLMs) due to: 1) lack of standardized benchmarks for traditional tasks, leading to inconsistent comparisons; 2) absence of benchmarks for mixed-modality generation, which fails to assess multimodal reasoning capabilities. We present a comprehensive evaluation framework designed to systematically assess U-MLLMs. Our benchmark includes: Standardized Traditional Task Evaluation. We sample from 12 datasets, covering 10 tasks with 30 subtasks, ensuring consistent and fair comparisons across studies." 2. Unified Task Assessment. We introduce five novel tasks testing multimodal reasoning, including image editing, commonsense QA with image generation, and geometric reasoning. 3. Comprehensive Model Benchmarking. We evaluate 12 leading U-MLLMs, such as Janus-Pro, EMU3, VILA-U, and Gemini2-flash, alongside specialized understanding (e.g., Claude-3.5-Sonnet) and generation models (e.g., DALL-E-3). Our findings reveal substantial performance gaps in existing U-MLLMs, highlighting the need for more robust models capable of handling mixed-modality tasks effectively. The code and evaluation data can be found in https://mme-unify.github.io/.
Authors: Ariadna Toh\`a-Dalmau (Department of Computer Science, Universitat Polit\`ecnica de Catalunya), Josep Rosin\'es-Fonoll (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona), Enrique Romero (Department of Computer Science, Universitat Polit\`ecnica de Catalunya, Intelligent Data Science and Artificial Intelligence Research Center), Ferran Mazzanti (Department of Physics, Universitat Polit\`ecnica de Catalunya), Ruben Martin-Pinardel (August Pi i Sunyer Biomedical Research Institute), Sonia Marias-Perez (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona), Carolina Bernal-Morales (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona, Diabetes Unit, Hospital Cl\'inic de Barcelona), Rafael Castro-Dominguez (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona), Andrea Mendez (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona), Emilio Ortega (August Pi i Sunyer Biomedical Research Institute, School of Medicine, Universitat de Barcelona), Irene Vinagre (August Pi i Sunyer Biomedical Research Institute, School of Medicine, Universitat de Barcelona), Marga Gimenez (August Pi i Sunyer Biomedical Research Institute, School of Medicine, Universitat de Barcelona), Alfredo Vellido (Department of Computer Science, Universitat Polit\`ecnica de Catalunya, Intelligent Data Science and Artificial Intelligence Research Center), Javier Zarranz-Ventura (Institut Cl\'inic d'Oftalmolog\'ia, Hospital Cl\'inic de Barcelona, School of Medicine, Universitat de Barcelona)
Abstract: This study aimed to develop a machine learning (ML) algorithm capable of determining cardiovascular risk in multimodal retinal images from patients with type 1 diabetes mellitus, distinguishing between moderate, high, and very high-risk levels. Radiomic features were extracted from fundus retinography, optical coherence tomography (OCT), and OCT angiography (OCTA) images. ML models were trained using these features either individually or combined with clinical data. A dataset of 597 eyes (359 individuals) was analyzed, and models trained only with radiomic features achieved AUC values of (0.79 $\pm$ 0.03) for identifying moderate risk cases from high and very high-risk cases, and (0.73 $\pm$ 0.07) for distinguishing between high and very high-risk cases. The addition of clinical variables improved all AUC values, reaching (0.99 $\pm$ 0.01) for identifying moderate risk cases and (0.95 $\pm$ 0.02) for differentiating between high and very high-risk cases. For very high CV risk, radiomics combined with OCT+OCTA metrics and ocular data achieved an AUC of (0.89 $\pm$ 0.02) without systemic data input. These results demonstrate that radiomic features obtained from multimodal retinal images are useful for discriminating and classifying CV risk labels, highlighting the potential of this oculomics approach for CV risk assessment.
Authors: Junchi Zhou, Haozhou Wang, Yoichiro Kato, Tejasri Nampally, P. Rajalakshmi, M. Balram, Keisuke Katsura, Hao Lu, Yue Mu, Wanneng Yang, Yangmingrui Gao, Feng Xiao, Hongtao Chen, Yuhao Chen, Wenjuan Li, Jingwen Wang, Fenghua Yu, Jian Zhou, Wensheng Wang, Xiaochun Hu, Yuanzhu Yang, Yanfeng Ding, Wei Guo, Shouyang Liu
Abstract: Developing computer vision-based rice phenotyping techniques is crucial for precision field management and accelerating breeding, thereby continuously advancing rice production. Among phenotyping tasks, distinguishing image components is a key prerequisite for characterizing plant growth and development at the organ scale, enabling deeper insights into eco-physiological processes. However, due to the fine structure of rice organs and complex illumination within the canopy, this task remains highly challenging, underscoring the need for a high-quality training dataset. Such datasets are scarce, both due to a lack of large, representative collections of rice field images and the time-intensive nature of annotation. To address this gap, we established the first comprehensive multi-class rice semantic segmentation dataset, RiceSEG. We gathered nearly 50,000 high-resolution, ground-based images from five major rice-growing countries (China, Japan, India, the Philippines, and Tanzania), encompassing over 6,000 genotypes across all growth stages. From these original images, 3,078 representative samples were selected and annotated with six classes (background, green vegetation, senescent vegetation, panicle, weeds, and duckweed) to form the RiceSEG dataset. Notably, the sub-dataset from China spans all major genotypes and rice-growing environments from the northeast to the south. Both state-of-the-art convolutional neural networks and transformer-based semantic segmentation models were used as baselines. While these models perform reasonably well in segmenting background and green vegetation, they face difficulties during the reproductive stage, when canopy structures are more complex and multiple classes are involved. These findings highlight the importance of our dataset for developing specialized segmentation models for rice and other crops.
Authors: Xiaoyu Tong, Zhi Zhang, Martha Lewis, Ekaterina Shutova
Abstract: Metaphor and humor share a lot of common ground, and metaphor is one of the most common humorous mechanisms. This study focuses on the humorous capacity of multimodal metaphors, which has not received due attention in the community. We take inspiration from the Incongruity Theory of humor, the Conceptual Metaphor Theory, and the annotation scheme behind the VU Amsterdam Metaphor Corpus, and developed a novel annotation scheme for humorous multimodal metaphor use in image-caption pairs. We create the Hummus Dataset of Humorous Multimodal Metaphor Use, providing expert annotation on 1k image-caption pairs sampled from the New Yorker Caption Contest corpus. Using the dataset, we test state-of-the-art multimodal large language models (MLLMs) on their ability to detect and understand humorous multimodal metaphor use. Our experiments show that current MLLMs still struggle with processing humorous multimodal metaphors, particularly with regard to integrating visual and textual information. We release our dataset and code at github.com/xiaoyuisrain/humorous-multimodal-metaphor-use.
Authors: Siqi Wang, Aoming Liu, Bryan A. Plummer
Abstract: Multi-source Domain Generalization (DG) aims to improve model robustness to new distributions. However, DG methods often overlook the effect of label noise, which can confuse a model during training, reducing performance. Limited prior work has analyzed DG method's noise-robustness, typically focused on an analysis of existing methods rather than new solutions. In this paper, we investigate this underexplored space, where models are evaluated under both distribution shifts and label noise, which we refer to as Noise-Aware Generalization (NAG). A natural solution to address label noise would be to combine a Learning with Noisy Labels (LNL) method with those from DG. Many LNL methods aim to detect distribution shifts in a class's samples, i.e., they assume that distribution shifts often correspond to label noise. However, in NAG distribution shifts can be due to label noise or domain shifts, breaking the assumptions used by LNL methods. A naive solution is to make a similar assumption made by many DG methods, where we presume to have domain labels during training, enabling us to isolate the two types of shifts. However, this ignores valuable cross-domain information. Specifically, our proposed DL4ND approach improves noise detection by taking advantage of the observation that noisy samples that may appear indistinguishable within a single domain often show greater variation when compared across domains. Experiments show that DL4ND significantly improves performance across four diverse datasets, offering a promising direction for tackling NAG.
Authors: Haozhan Tang, Tianyi Zhang, Oliver Kroemer, Matthew Johnson-Roberson, Weiming Zhi
Abstract: Robots operating in unstructured environments often require accurate and consistent object-level representations. This typically requires segmenting individual objects from the robot's surroundings. While recent large models such as Segment Anything (SAM) offer strong performance in 2D image segmentation. These advances do not translate directly to performance in the physical 3D world, where they often over-segment objects and fail to produce consistent mask correspondences across views. In this paper, we present GraphSeg, a framework for generating consistent 3D object segmentations from a sparse set of 2D images of the environment without any depth information. GraphSeg adds edges to graphs and constructs dual correspondence graphs: one from 2D pixel-level similarities and one from inferred 3D structure. We formulate segmentation as a problem of edge addition, then subsequent graph contraction, which merges multiple 2D masks into unified object-level segmentations. We can then leverage \emph{3D foundation models} to produce segmented 3D representations. GraphSeg achieves robust segmentation with significantly fewer images and greater accuracy than prior methods. We demonstrate state-of-the-art performance on tabletop scenes and show that GraphSeg enables improved performance on downstream robotic manipulation tasks. Code available at https://github.com/tomtang502/graphseg.git.
Authors: Fatemeh Javadian, Zahra Aminparast, Johannes Stegmaier, Abin Jose
Abstract: This study explores the application of supervised and unsupervised autoencoders (AEs) to automate nuclei classification in clear cell renal cell carcinoma (ccRCC) images, a diagnostic task traditionally reliant on subjective visual grading by pathologists. We evaluate various AE architectures, including standard AEs, contractive AEs (CAEs), and discriminative AEs (DAEs), as well as a classifier-based discriminative AE (CDAE), optimized using the hyperparameter tuning tool Optuna. Bhattacharyya distance is selected from several metrics to assess class separability in the latent space, revealing challenges in distinguishing adjacent grades using unsupervised models. CDAE, integrating a supervised classifier branch, demonstrated superior performance in both latent space separation and classification accuracy. Given that CDAE-CNN achieved notable improvements in classification metrics, affirming the value of supervised learning for class-specific feature extraction, F1 score was incorporated into the tuning process to optimize classification performance. Results show significant improvements in identifying aggressive ccRCC grades by leveraging the classification capability of AE through latent clustering followed by fine-grained classification. Our model outperforms the current state of the art, CHR-Network, across all evaluated metrics. These findings suggest that integrating a classifier branch in AEs, combined with neural architecture search and contrastive learning, enhances grading automation in ccRCC pathology, particularly in detecting aggressive tumor grades, and may improve diagnostic accuracy.
Authors: Akis Nousias, Efklidis Katsaros, Evangelos Syrmos, Panagiotis Radoglou-Grammatikis, Thomas Lagkas, Vasileios Argyriou, Ioannis Moscholios, Evangelos Markakis, Sotirios Goudos, Panagiotis Sarigiannidis
Abstract: Malware detection is increasingly challenged by evolving techniques like obfuscation and polymorphism, limiting the effectiveness of traditional methods. Meanwhile, the widespread adoption of software containers has introduced new security challenges, including the growing threat of malicious software injection, where a container, once compromised, can serve as entry point for further cyberattacks. In this work, we address these security issues by introducing a method to identify compromised containers through machine learning analysis of their file systems. We cast the entire software containers into large RGB images via their tarball representations, and propose to use established Convolutional Neural Network architectures on a streaming, patch-based manner. To support our experiments, we release the COSOCO dataset--the first of its kind--containing 3364 large-scale RGB images of benign and compromised software containers at https://huggingface.co/datasets/k3ylabs/cosoco-image-dataset. Our method detects more malware and achieves higher F1 and Recall scores than all individual and ensembles of VirusTotal engines, demonstrating its effectiveness and setting a new standard for identifying malware-compromised software containers.
URLs: https://huggingface.co/datasets/k3ylabs/cosoco-image-dataset.
Authors: Chen Hu, Enrica Tricomi, Eojin Rho, Daekyum Kim, Lorenzo Masia, Shan Luo, Letizia Gionfrida
Abstract: Grasping is a fundamental skill for interacting with and manipulating objects in the environment. However, this ability can be challenging for individuals with hand impairments. Soft hand exoskeletons designed to assist grasping can enhance or restore essential hand functions, yet controlling these soft exoskeletons to support users effectively remains difficult due to the complexity of understanding the environment. This study presents a vision-based predictive control framework that leverages contextual awareness from depth perception to predict the grasping target and determine the next control state for activation. Unlike data-driven approaches that require extensive labelled datasets and struggle with generalizability, our method is grounded in geometric modelling, enabling robust adaptation across diverse grasping scenarios. The Grasping Ability Score (GAS) was used to evaluate performance, with our system achieving a state-of-the-art GAS of 91% across 15 objects and healthy participants, demonstrating its effectiveness across different object types. The proposed approach maintained reconstruction success for unseen objects, underscoring its enhanced generalizability compared to learning-based models.
Authors: Zhe Wang, Yifei Zhu
Abstract: Neural Radiance Fields (NeRF) is a cutting-edge neural network-based technique for novel view synthesis in 3D reconstruction. However, its significant computational demands pose challenges for deployment on mobile devices. While mesh-based NeRF solutions have shown potential in achieving real-time rendering on mobile platforms, they often fail to deliver high-quality reconstructions when rendering practical complex scenes. Additionally, the non-negligible memory overhead caused by pre-computed intermediate results complicates their practical application. To overcome these challenges, we present NeRFlex, a resource-aware, high-resolution, real-time rendering framework for complex scenes on mobile devices. NeRFlex integrates mobile NeRF rendering with multi-NeRF representations that decompose a scene into multiple sub-scenes, each represented by an individual NeRF network. Crucially, NeRFlex considers both memory and computation constraints as first-class citizens and redesigns the reconstruction process accordingly. NeRFlex first designs a detail-oriented segmentation module to identify sub-scenes with high-frequency details. For each NeRF network, a lightweight profiler, built on domain knowledge, is used to accurately map configurations to visual quality and memory usage. Based on these insights and the resource constraints on mobile devices, NeRFlex presents a dynamic programming algorithm to efficiently determine configurations for all NeRF representations, despite the NP-hardness of the original decision problem. Extensive experiments on real-world datasets and mobile devices demonstrate that NeRFlex achieves real-time, high-quality rendering on commercial mobile devices.
Authors: Gianluca Maselli, Vieri Giuliano Santucci
Abstract: Dealing with environments with sparse rewards has always been crucial for systems developed to operate in autonomous open-ended learning settings. Intrinsic Motivations could be an effective way to help Deep Reinforcement Learning algorithms learn in such scenarios. In fact, intrinsic reward signals, such as novelty or curiosity, are generally adopted to improve exploration when extrinsic rewards are delayed or absent. Building on previous works, we tackle the problem of learning policies in the presence of sparse rewards by proposing a two-level architecture that alternates an ''intrinsically driven'' phase of exploration and autonomous sub-goal generation, to a phase of sparse reward, goal-directed policy learning. The idea is to build several small networks, each one specialized on a particular sub-path, and use them as starting points for future exploration without the need to further explore from scratch previously learnt paths. Two versions of the system have been trained and tested in the Gym SuperMarioBros environment without considering any additional extrinsic reward. The results show the validity of our approach and the importance of autonomously segment the environment to generate an efficient path towards the final goal.
Authors: Mohammad Reza Yousefi, Ali Bakrani, Amin Dehghani
Abstract: Diabetic Retinopathy (DR) is a serious and common complication of diabetes, caused by prolonged high blood sugar levels that damage the small retinal blood vessels. If left untreated, DR can progress to retinal vein occlusion and stimulate abnormal blood vessel growth, significantly increasing the risk of blindness. Traditional diabetes diagnosis methods often utilize convolutional neural networks (CNNs) to extract visual features from retinal images, followed by classification algorithms such as decision trees and k-nearest neighbors (KNN) for disease detection. However, these approaches face several challenges, including low accuracy and sensitivity, lengthy machine learning (ML) model training due to high data complexity and volume, and the use of limited datasets for testing and evaluation. This study investigates the application of transfer learning (TL) to enhance ML model performance in DR detection. Key improvements include dimensionality reduction, optimized learning rate adjustments, and advanced parameter tuning algorithms, aimed at increasing efficiency and diagnostic accuracy. The proposed model achieved an overall accuracy of 84% on the testing dataset, outperforming prior studies. The highest class-specific accuracy reached 89%, with a maximum sensitivity of 97% and an F1-score of 92%, demonstrating strong performance in identifying DR cases. These findings suggest that TL-based DR screening is a promising approach for early diagnosis, enabling timely interventions to prevent vision loss and improve patient outcomes.
Authors: Spyros Kondylatos, Nikolaos Ioannis Bountos, Ioannis Prapas, Angelos Zavras, Gustau Camps-Valls, Ioannis Papoutsis
Abstract: Label noise poses a significant challenge in Earth Observation (EO), often degrading the performance and reliability of supervised Machine Learning (ML) models. Yet, given the critical nature of several EO applications, developing robust and trustworthy ML solutions is essential. In this study, we take a step in this direction by leveraging probabilistic ML to model input-dependent label noise and quantify data uncertainty in EO tasks, accounting for the unique noise sources inherent in the domain. We train uncertainty-aware probabilistic models across a broad range of high-impact EO applications-spanning diverse noise sources, input modalities, and ML configurations-and introduce a dedicated pipeline to assess their accuracy and reliability. Our experimental results show that the uncertainty-aware models consistently outperform the standard deterministic approaches across most datasets and evaluation metrics. Moreover, through rigorous uncertainty evaluation, we validate the reliability of the predicted uncertainty estimates, enhancing the interpretability of model predictions. Our findings emphasize the importance of modeling label noise and incorporating uncertainty quantification in EO, paving the way for more accurate, reliable, and trustworthy ML solutions in the field.
Authors: Shuofei Qiao, Zhisong Qiu, Baochang Ren, Xiaobin Wang, Xiangyuan Ru, Ningyu Zhang, Xiang Chen, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen
Abstract: Large Language Models (LLMs) have achieved considerable performance across various agentic planning tasks. However, traditional agent planning approaches adopt a "flood irrigation" methodology that indiscriminately injects gold trajectories, external feedback, and domain knowledge into agent models. This practice overlooks the fundamental human cognitive principle of situational self-awareness during decision-making-the ability to dynamically assess situational demands and strategically employ resources during decision-making. We propose agentic knowledgeable self-awareness to address this gap, a novel paradigm enabling LLM-based agents to autonomously regulate knowledge utilization. Specifically, we propose KnowSelf, a data-centric approach that applies agents with knowledgeable self-awareness like humans. Concretely, we devise a heuristic situation judgement criterion to mark special tokens on the agent's self-explored trajectories for collecting training data. Through a two-stage training process, the agent model can switch between different situations by generating specific special tokens, achieving optimal planning effects with minimal costs. Our experiments demonstrate that KnowSelf can outperform various strong baselines on different tasks and models with minimal use of external knowledge. Code is available at https://github.com/zjunlp/KnowSelf.
Authors: Runnan Fang, Xiaobin Wang, Yuan Liang, Shuofei Qiao, Jialong Wu, Zekun Xi, Ningyu Zhang, Yong Jiang, Pengjun Xie, Fei Huang, Huajun Chen
Abstract: In the interaction between agents and their environments, agents expand their capabilities by planning and executing actions. However, LLM-based agents face substantial challenges when deployed in novel environments or required to navigate unconventional action spaces. To empower agents to autonomously explore environments, optimize workflows, and enhance their understanding of actions, we propose SynWorld, a framework that allows agents to synthesize possible scenarios with multi-step action invocation within the action space and perform Monte Carlo Tree Search (MCTS) exploration to effectively refine their action knowledge in the current environment. Our experiments demonstrate that SynWorld is an effective and general approach to learning action knowledge in new environments. Code is available at https://github.com/zjunlp/SynWorld.
Authors: Badhan Kumar Das, Gengyan Zhao, Han Liu, Thomas J. Re, Dorin Comaniciu, Eli Gibson, Andreas Maier
Abstract: Pretrain techniques, whether supervised or self-supervised, are widely used in deep learning to enhance model performance. In real-world clinical scenarios, different sets of magnetic resonance (MR) contrasts are often acquired for different subjects/cases, creating challenges for deep learning models assuming consistent input modalities among all the cases and between pretrain and finetune. Existing methods struggle to maintain performance when there is an input modality/contrast set mismatch with the pretrained model, often resulting in degraded accuracy. We propose an adaptive Vision Transformer (AdaViT) framework capable of handling variable set of input modalities for each case. We utilize a dynamic tokenizer to encode different input image modalities to tokens and take advantage of the characteristics of the transformer to build attention mechanism across variable length of tokens. Through extensive experiments, we demonstrate that this architecture effectively transfers supervised pretrained models to new datasets with different input modality/contrast sets, resulting in superior performance on zero-shot testing, few-shot finetuning, and backward transferring in brain infarct and brain tumor segmentation tasks. Additionally, for self-supervised pretrain, the proposed method is able to maximize the pretrain data and facilitate transferring to diverse downstream tasks with variable sets of input modalities.
Authors: Jun Ma, Zongxin Yang, Sumin Kim, Bihui Chen, Mohammed Baharoon, Adibvafa Fallahpour, Reza Asakereh, Hongwei Lyu, Bo Wang
Abstract: Medical image and video segmentation is a critical task for precision medicine, which has witnessed considerable progress in developing task or modality-specific and generalist models for 2D images. However, there have been limited studies on building general-purpose models for 3D images and videos with comprehensive user studies. Here, we present MedSAM2, a promptable segmentation foundation model for 3D image and video segmentation. The model is developed by fine-tuning the Segment Anything Model 2 on a large medical dataset with over 455,000 3D image-mask pairs and 76,000 frames, outperforming previous models across a wide range of organs, lesions, and imaging modalities. Furthermore, we implement a human-in-the-loop pipeline to facilitate the creation of large-scale datasets resulting in, to the best of our knowledge, the most extensive user study to date, involving the annotation of 5,000 CT lesions, 3,984 liver MRI lesions, and 251,550 echocardiogram video frames, demonstrating that MedSAM2 can reduce manual costs by more than 85%. MedSAM2 is also integrated into widely used platforms with user-friendly interfaces for local and cloud deployment, making it a practical tool for supporting efficient, scalable, and high-quality segmentation in both research and healthcare environments.
Authors: Kate Sanders, Benjamin Van Durme
Abstract: To develop general-purpose collaborative agents, humans need reliable AI systems that can (1) adapt to new domains and (2) transparently reason with uncertainty to allow for verification and correction. Black-box models demonstrate powerful data processing abilities but do not satisfy these criteria due to their opaqueness, domain specificity, and lack of uncertainty awareness. We introduce Bonsai, a compositional and probabilistic reasoning system that generates adaptable inference trees by retrieving relevant grounding evidence and using it to compute likelihoods of sub-claims derived from broader natural language inferences. Bonsai's reasoning power is tunable at test-time via evidence scaling and it demonstrates reliable handling of varied domains including transcripts, photographs, videos, audio, and databases. Question-answering and human alignment experiments demonstrate that Bonsai matches the performance of domain-specific black-box methods while generating interpretable, grounded, and uncertainty-aware reasoning traces.
Authors: Qiangqiang Wu, Tianyu Yang, Ziquan Liu, Wei Lin, Baoyuan Wu, Antoni B. Chan
Abstract: This paper studies masked autoencoder (MAE) video pre-training for various temporal matching-based downstream tasks, i.e., object-level tracking tasks including video object tracking (VOT) and video object segmentation (VOS), self-supervised visual correspondence learning, dense tracking tasks including optical flow estimation and long-term point tracking, and 3D point cloud tracking. Specifically, our work explores to provide a general representation to boost the temporal matching ability in various downstream tracking tasks. To achieve this, we firstly find that a simple extension of MAE, which randomly masks out frame patches in videos and reconstruct the frame pixels, heavily relies on spatial cues while ignoring temporal relations for frame reconstruction, thus leading to sub-optimal temporal matching representations. To alleviate this, we propose DropMAE, which adaptively performs spatial-attention dropout in the frame reconstruction to facilitate temporal correspondence learning in videos. We obtain several important findings with DropMAE: 1) DropMAE is a strong and efficient temporal matching learner, which achieves better fine-tuning results on matching-based tasks than the ImageNet-based MAE with 2x faster pre-training speed. 2) DropMAE is effective for different tracking tasks, i.e., object-level matching tasks including VOT and VOS, dense tracking tasks including optical flow estimation and tracking any point (TAP), and even 3D tracking in the different modality of point cloud data. Since none exists, we build ViT-based trackers for different downstream tracking tasks, and our pre-trained DropMAE model can be directly loaded in these ViT-based trackers for fine-tuning without further modifications. Experiments on 6 downstream tracking tasks demonstrate the effectiveness of DropMAE as a general pre-trained representation for diverse tracking tasks.
Authors: Anbang Yang, Ge Jin, Junjie Huang, Yao Wang, John-Ross Rizzo, Chen Feng
Abstract: Real-time visual localization often utilizes online computing, for which query images or videos are transmitted to remote servers for visual place recognition (VPR). However, limited network bandwidth necessitates image-quality reduction and thus the degradation of global image descriptors, reducing VPR accuracy. We address this issue at the descriptor extraction level with a knowledge-distillation methodology that learns feature representations from high-quality images to extract more discriminative descriptors from low-quality images. Our approach includes the Inter-channel Correlation Knowledge Distillation (ICKD) loss, Mean Squared Error (MSE) loss, and Triplet loss. We validate the proposed losses on multiple VPR methods and datasets subjected to JPEG compression, resolution reduction, and video quantization. We obtain significant improvements in VPR recall rates under all three tested modalities of lowered image quality. Furthermore, we fill a gap in VPR literature on video-based data and its influence on VPR performance. This work contributes to more reliable place recognition in resource-constrained environments.
Authors: Mingzhao Yang, Shangchao Su, Bin Li, Xiangyang Xue
Abstract: In recent years, One-shot Federated Learning methods based on Diffusion Models have garnered increasing attention due to their remarkable performance. However, most of these methods require the deployment of foundation models on client devices, which significantly raises the computational requirements and reduces their adaptability to heterogeneous client models compared to traditional FL methods. In this paper, we propose FedLMG, a heterogeneous one-shot Federated learning method with Local Model-Guided diffusion models. Briefly speaking, in FedLMG, clients do not need access to any foundation models but only train and upload their local models, which is consistent with traditional FL methods. On the clients, we employ classification loss and BN loss to capture the broad category features and detailed contextual features of the client distributions. On the server, based on the uploaded client models, we utilize backpropagation to guide the server's DM in generating synthetic datasets that comply with the client distributions, which are then used to train the aggregated model. By using the locally trained client models as a medium to transfer client knowledge, our method significantly reduces the computational requirements on client devices and effectively adapts to scenarios with heterogeneous clients. Extensive quantitation and visualization experiments on three large-scale real-world datasets, along with theoretical analysis, demonstrate that the synthetic datasets generated by FedLMG exhibit comparable quality and diversity to the client datasets, which leads to an aggregated model that outperforms all compared methods and even the performance ceiling, further elucidating the significant potential of utilizing DMs in FL.
Authors: Zitian Tang, Rohan Myer Krishnan, Zhiqiu Yu, Chen Sun
Abstract: Learning from (procedural) videos has increasingly served as a pathway for embodied agents to acquire skills from human demonstrations. To do this, video understanding models must be able to obtain structured understandings, such as the temporal segmentation of a demonstration into sequences of actions and skills, and to generalize the understandings to novel environments, tasks, and problem domains. In pursuit of this goal, we introduce Spacewalk-18, a benchmark containing two tasks: (1) step recognition and (2) video question answering, over a dataset of temporally segmented and labeled tasks in International Space Station spacewalk recordings. In tandem, the two tasks quantify a model's ability to: (1) generalize to novel domains; (2) utilize long temporal context and multimodal (e.g. visual and speech) information. Our extensive experimental analysis highlights the challenges of Spacewalk-18, but also suggests best practices for domain generalization and long-form understanding. Notably, we discover a promising adaptation via summarization technique that leads to significant performance improvement without model fine-tuning. The Spacewalk-18 benchmark is released at https://brown-palm.github.io/Spacewalk-18/.
Authors: Ibrahim Ethem Hamamci, Sezgin Er, Chenyu Wang, Furkan Almas, Ayse Gulnihan Simsek, Sevval Nil Esirgun, Irem Doga, Omer Faruk Durugol, Weicheng Dai, Murong Xu, Muhammed Furkan Dasdelen, Bastian Wittmann, Tamaz Amiranashvili, Enis Simsar, Mehmet Simsar, Emine Bensu Erdemir, Abdullah Alanbay, Anjany Sekuboyina, Berkan Lafci, Christian Bluethgen, Kayhan Batmanghelich, Mehmet Kemal Ozdemir, Bjoern Menze
Abstract: While computer vision has achieved tremendous success with multimodal encoding and direct textual interaction with images via chat-based large language models, similar advancements in medical imaging AI, particularly in 3D imaging, have been limited due to the scarcity of comprehensive datasets. To address this critical gap, we introduce CT-RATE, the first dataset that pairs 3D medical images with corresponding textual reports. CT-RATE comprises 25,692 non-contrast 3D chest CT scans from 21,304 unique patients. Through various reconstructions, these scans are expanded to 50,188 volumes, totaling over 14.3 million 2D slices. Each scan is accompanied by its corresponding radiology report. Leveraging CT-RATE, we develop CT-CLIP, a CT-focused contrastive language-image pretraining framework designed for broad applications without the need for task-specific training. We demonstrate how CT-CLIP can be used in two tasks: multi-abnormality detection and case retrieval. Remarkably, in multi-abnormality detection, CT-CLIP outperforms state-of-the-art fully supervised models across all key metrics, effectively eliminating the need for manual annotation. In case retrieval, it efficiently retrieves relevant cases using either image or textual queries, thereby enhancing knowledge dissemination. By combining CT-CLIP's vision encoder with a pretrained large language model, we create CT-CHAT, a vision-language foundational chat model for 3D chest CT volumes. Finetuned on over 2.7 million question-answer pairs derived from the CT-RATE dataset, CT-CHAT surpasses other multimodal AI assistants, underscoring the necessity for specialized methods in 3D medical imaging. Collectively, the open-source release of CT-RATE, CT-CLIP, and CT-CHAT not only addresses critical challenges in 3D medical imaging, but also lays the groundwork for future innovations in medical AI and improved patient care.
Authors: Heemin Yang, Jaesung Rim, Seungyong Lee, Seung-Hwan Baek, Sunghyun Cho
Abstract: In this paper, we present GyroDeblurNet, a novel single-image deblurring method that utilizes a gyro sensor to resolve the ill-posedness of image deblurring. The gyro sensor provides valuable information about camera motion that can improve deblurring quality. However, exploiting real-world gyro data is challenging due to errors from various sources. To handle these errors, GyroDeblurNet is equipped with two novel neural network blocks: a gyro refinement block and a gyro deblurring block. The gyro refinement block refines the erroneous gyro data using the blur information from the input image. The gyro deblurring block removes blur from the input image using the refined gyro data and further compensates for gyro error by leveraging the blur information from the input image. For training a neural network with erroneous gyro data, we propose a training strategy based on the curriculum learning. We also introduce a novel gyro data embedding scheme to represent real-world intricate camera shakes. Finally, we present both synthetic and real-world datasets for training and evaluating gyro-based single image deblurring. Our experiments demonstrate that our approach achieves state-of-the-art deblurring quality by effectively utilizing erroneous gyro data.
Authors: Sinisa Stekovic, Arslan Artykov, Stefan Ainetter, Mattia D'Urso, Friedrich Fraundorfer
Abstract: We propose PyTorchGeoNodes, a differentiable module for reconstructing 3D objects and their parameters from images using interpretable shape programs. Unlike traditional CAD model retrieval, shape programs allow reasoning about semantic parameters, editing, and a low memory footprint. Despite their potential, shape programs for 3D scene understanding have been largely overlooked. Our key contribution is enabling gradient-based optimization by parsing shape programs, or more precisely procedural models designed in Blender, into efficient PyTorch code. While there are many possible applications of our PyTochGeoNodes, we show that a combination of PyTorchGeoNodes with genetic algorithm is a method of choice to optimize both discrete and continuous shape program parameters for 3D reconstruction and understanding of 3D object parameters. Our modular framework can be further integrated with other reconstruction algorithms, and we demonstrate one such integration to enable procedural Gaussian splatting. Our experiments on the ScanNet dataset show that our method achieves accurate reconstructions while enabling, until now, unseen level of 3D scene understanding.
Authors: Georgia Sovatzidi, Michael D. Vasilakakis, Dimitris K. Iakovidis
Abstract: Several deep learning (DL) approaches have been proposed to deal with image classification tasks. However, despite their effectiveness, they lack interpretability, as they are unable to explain or justify their results. To address the challenge of interpretable image classification, this paper introduces a novel framework, named Interpretable Intuitionistic Fuzzy Cognitive Maps (I2FCMs).Intuitionistic FCMs (iFCMs) have been proposed as an extension of FCMs offering a natural mechanism to assess the quality of their output through the estimation of hesitancy, a concept resembling human hesitation in decision making. In the context of image classification, hesitancy is considered as a degree of unconfidence with which an image is categorized to a class. To the best of our knowledge this is the first time iFCMs are applied for image classification. Further novel contributions of the introduced framework include the following: a) a feature extraction process focusing on the most informative image regions; b) a learning algorithm for automatic data-driven determination of the intuitionistic fuzzy interconnections of the iFCM, thereby reducing human intervention in the definition of the graph structure; c) an inherently interpretable classification approach based on image contents, providing understandable explanations of its predictions, using linguistic terms. Furthermore, the proposed I2FCM framework can be applied to DL models, including Convolutional Neural Network (CNN), rendering them interpretable. The effectiveness of I2FCM is evaluated on publicly available datasets, and the results confirm that it can provide enhanced classification performance, while providing interpretable inferences.
Authors: Jianwen Jiang, Chao Liang, Jiaqi Yang, Gaojie Lin, Tianyun Zhong, Yanbo Zheng
Abstract: With the introduction of diffusion-based video generation techniques, audio-conditioned human video generation has recently achieved significant breakthroughs in both the naturalness of motion and the synthesis of portrait details. Due to the limited control of audio signals in driving human motion, existing methods often add auxiliary spatial signals to stabilize movements, which may compromise the naturalness and freedom of motion. In this paper, we propose an end-to-end audio-only conditioned video diffusion model named Loopy. Specifically, we designed an inter- and intra-clip temporal module and an audio-to-latents module, enabling the model to leverage long-term motion information from the data to learn natural motion patterns and improving audio-portrait movement correlation. This method removes the need for manually specified spatial motion templates used in existing methods to constrain motion during inference. Extensive experiments show that Loopy outperforms recent audio-driven portrait diffusion models, delivering more lifelike and high-quality results across various scenarios.
Authors: Hayeon Jo, Hyesong Choi, Minhee Cho, Dongbo Min
Abstract: Transfer learning based on full fine-tuning (FFT) of the pre-trained encoder and task-specific decoder becomes increasingly complex as deep models grow exponentially. Parameter efficient fine-tuning (PEFT) approaches using adapters consisting of small learnable layers have emerged as an alternative to FFT, achieving comparable performance while maintaining high training efficiency. However, the inflexibility of the adapter with respect to input instances limits its capability of learning task-specific information in diverse downstream tasks. In this paper, we propose a novel PEFT approach, input-Conditioned transFormer, termed iConFormer, that leverages a dynamic adapter conditioned on the input instances. To secure flexible learning ability on input instances in various downstream tasks, we introduce an input-Conditioned Network (iCoN) in the dynamic adapter that enables instance-level feature transformation. To be specific, iCoN generates channel-wise convolutional kernels for each feature and transform it using adaptive convolution process to effectively capture task-specific and fine-grained details tailor to downstream tasks. Experimental results demonstrate that by tuning just 1.6% to 2.8% of the Transformer backbone parameters, iConFormer achieves performance comparable to FFT in monocular depth estimation and semantic segmentation, while outperforming it in image classification and instance segmentation. Also, the proposed method consistently outperforms recent PEFT methods for all the tasks mentioned above.
Authors: Wenfeng Huang, Guoan Xu, Wenjing Jia, Stuart Perry, Guangwei Gao
Abstract: Images captured in challenging environments--such as nighttime, smoke, rainy weather, and underwater--often suffer from significant degradation, resulting in a substantial loss of visual quality. The effective restoration of these degraded images is critical for the subsequent vision tasks. While many existing approaches have successfully incorporated specific priors for individual tasks, these tailored solutions limit their applicability to other degradations. In this work, we propose a universal network architecture, dubbed ``ReviveDiff'', which can address various degradations and bring images back to life by enhancing and restoring their quality. Our approach is inspired by the observation that, unlike degradation caused by movement or electronic issues, quality degradation under adverse conditions primarily stems from natural media (such as fog, water, and low luminance), which generally preserves the original structures of objects. To restore the quality of such images, we leveraged the latest advancements in diffusion models and developed ReviveDiff to restore image quality from both macro and micro levels across some key factors determining image quality, such as sharpness, distortion, noise level, dynamic range, and color accuracy. We rigorously evaluated ReviveDiff on seven benchmark datasets covering five types of degrading conditions: Rainy, Underwater, Low-light, Smoke, and Nighttime Hazy. Our experimental results demonstrate that ReviveDiff outperforms the state-of-the-art methods both quantitatively and visually.
Authors: Xingyu Ji, Shenghai Yuan, Jianping Li, Pengyu Yin, Haozhi Cao, Lihua Xie
Abstract: LiDAR bundle adjustment (BA) is an effective approach to reduce the drifts in pose estimation from the front-end. Existing works on LiDAR BA usually rely on predefined geometric features for landmark representation. This reliance restricts generalizability, as the system will inevitably deteriorate in environments where these specific features are absent. To address this issue, we propose SGBA, a LiDAR BA scheme that models the environment as a semantic Gaussian mixture model (GMM) without predefined feature types. This approach encodes both geometric and semantic information, offering a comprehensive and general representation adaptable to various environments. Additionally, to limit computational complexity while ensuring generalizability, we propose an adaptive semantic selection framework that selects the most informative semantic clusters for optimization by evaluating the condition number of the cost function. Lastly, we introduce a probabilistic feature association scheme that considers the entire probability density of assignments, which can manage uncertainties in measurement and initial pose estimation. We have conducted various experiments and the results demonstrate that SGBA can achieve accurate and robust pose refinement even in challenging scenarios with low-quality initial pose estimation and limited geometric features. We plan to open-source the work for the benefit of the community https://github.com/Ji1Xinyu/SGBA.
Authors: Soobin Um, Jong Chul Ye
Abstract: We investigate the generation of minority samples using pretrained text-to-image (T2I) latent diffusion models. Minority instances, in the context of T2I generation, can be defined as ones living on low-density regions of text-conditional data distributions. They are valuable for various applications of modern T2I generators, such as data augmentation and creative AI. Unfortunately, existing pretrained T2I diffusion models primarily focus on high-density regions, largely due to the influence of guided samplers (like CFG) that are essential for high-quality generation. To address this, we present a novel framework to counter the high-density-focus of T2I diffusion models. Specifically, we first develop an online prompt optimization framework that encourages emergence of desired properties during inference while preserving semantic contents of user-provided prompts. We subsequently tailor this generic prompt optimizer into a specialized solver that promotes generation of minority features by incorporating a carefully-crafted likelihood objective. Extensive experiments conducted across various types of T2I models demonstrate that our approach significantly enhances the capability to produce high-quality minority instances compared to existing samplers. Code is available at https://github.com/soobin-um/MinorityPrompt.
Authors: Tassilo Wald, Constantin Ulrich, Stanislav Lukyanenko, Andrei Goncharov, Alberto Paderno, Maximilian Miller, Leander Maerkisch, Paul F. J\"ager, Klaus Maier-Hein
Abstract: Self-Supervised Learning (SSL) presents an exciting opportunity to unlock the potential of vast, untapped clinical datasets, for various downstream applications that suffer from the scarcity of labeled data. While SSL has revolutionized fields like natural language processing and computer vision, its adoption in 3D medical image computing has been limited by three key pitfalls: Small pre-training dataset sizes, architectures inadequate for 3D medical image analysis, and insufficient evaluation practices. In this paper, we address these issues by i) leveraging a large-scale dataset of 39k 3D brain MRI volumes and ii) using a Residual Encoder U-Net architecture within the state-of-the-art nnU-Net framework. iii) A robust development framework, incorporating 5 development and 8 testing brain MRI segmentation datasets, allowed performance-driven design decisions to optimize the simple concept of Masked Auto Encoders (MAEs) for 3D CNNs. The resulting model not only surpasses previous SSL methods but also outperforms the strong nnU-Net baseline by an average of approximately 3 Dice points setting a new state-of-the-art. Our code and models are made available here.
Authors: Huancheng Chen, Jingtao Li, Nidham Gazagnadou, Weiming Zhuang, Chen Chen, Lingjuan Lyu
Abstract: In the era of foundation models, we revisit continual learning~(CL), which aims to enable vision transformers (ViTs) to learn new tasks over time. However, as the scale of these models increases, catastrophic forgetting remains a persistent challenge, particularly in the presence of significant domain shifts across tasks. Recent studies highlight a crossover between CL techniques and parameter-efficient fine-tuning (PEFT), which focuses on fine-tuning only a small set of trainable parameters to adapt to downstream tasks, such as low-rank adaptation (LoRA). While LoRA achieves faster convergence and requires fewer trainable parameters, it has seldom been explored in the context of continual learning. To address this gap, we propose a novel PEFT-CL method called Dual Low-Rank Adaptation (DualLoRA), which introduces both an orthogonal LoRA adapter and a residual LoRA adapter parallel to pre-trained weights in each layer. These components are orchestrated by a dynamic memory mechanism to strike a balance between stability and plasticity. The orthogonal LoRA adapter's parameters are updated in an orthogonal subspace of previous tasks to mitigate catastrophic forgetting, while the residual LoRA adapter's parameters are updated in the residual subspace spanned by task-specific bases without interaction across tasks, offering complementary capabilities for fine-tuning new tasks. On ViT-based models, we demonstrate that DualLoRA offers significant advantages in accuracy, inference speed, and memory efficiency over existing CL methods across multiple benchmarks.
Authors: Chong Wang, Fengbei Liu, Yuanhong Chen, Helen Frazer, Gustavo Carneiro
Abstract: Recent advances in prototypical learning have shown remarkable potential to provide useful decision interpretations associating activation maps and predictions with class-specific training prototypes. Such prototypical learning has been well-studied for various single-label diseases, but for quite relevant and more challenging multi-label diagnosis, where multiple diseases are often concurrent within an image, existing prototypical learning models struggle to obtain meaningful activation maps and effective class prototypes due to the entanglement of the multiple diseases. In this paper, we present a novel Cross- and Intra-image Prototypical Learning (CIPL) framework, for accurate multi-label disease diagnosis and interpretation from medical images. CIPL takes advantage of common cross-image semantics to disentangle the multiple diseases when learning the prototypes, allowing a comprehensive understanding of complicated pathological lesions. Furthermore, we propose a new two-level alignment-based regularisation strategy that effectively leverages consistent intra-image information to enhance interpretation robustness and predictive performance. Extensive experiments show that our CIPL attains the state-of-the-art (SOTA) classification accuracy in two public multi-label benchmarks of disease diagnosis: thoracic radiography and fundus images. Quantitative interpretability results show that CIPL also has superiority in weakly-supervised thoracic disease localisation over other leading saliency- and prototype-based explanation methods.
Authors: Jiaxuan Chen, Bo Zhang, Qingdong He, Jinlong Peng, Li Niu
Abstract: Generative image composition aims to regenerate the given foreground object in the background image to produce a realistic composite image. The existing methods are struggling to preserve the foreground details and adjust the foreground pose/viewpoint at the same time. In this work, we propose an effective finetuning strategy for generative image composition model, in which we finetune a pretrained model using one or more images containing the same foreground object. Moreover, we propose a multi-reference strategy, which allows the model to take in multiple reference images of the foreground object. The experiments on MureCOM dataset verify the effectiveness of our method. The code and model have been released at https://github.com/bcmi/MureObjectStitch-Image-Composition.
URLs: https://github.com/bcmi/MureObjectStitch-Image-Composition.
Authors: Cheng Jin, Luyang Luo, Huangjing Lin, Jun Hou, Hao Chen
Abstract: Fine-grained classification of whole slide images (WSIs) is essential in precision oncology, enabling precise cancer diagnosis and personalized treatment strategies. The core of this task involves distinguishing subtle morphological variations within the same broad category of gigapixel-resolution images, which presents a significant challenge. While the multi-instance learning (MIL) paradigm alleviates the computational burden of WSIs, existing MIL methods often overlook hierarchical label correlations, treating fine-grained classification as a flat multi-class classification task. To overcome these limitations, we introduce a novel hierarchical multi-instance learning (HMIL) framework. By facilitating on the hierarchical alignment of inherent relationships between different hierarchy of labels at instance and bag level, our approach provides a more structured and informative learning process. Specifically, HMIL incorporates a class-wise attention mechanism that aligns hierarchical information at both the instance and bag levels. Furthermore, we introduce supervised contrastive learning to enhance the discriminative capability for fine-grained classification and a curriculum-based dynamic weighting module to adaptively balance the hierarchical feature during training. Extensive experiments on our large-scale cytology cervical cancer (CCC) dataset and two public histology datasets, BRACS and PANDA, demonstrate the state-of-the-art class-wise and overall performance of our HMIL framework. Our source code is available at https://github.com/ChengJin-git/HMIL.
Authors: Andong Deng, Tongjia Chen, Shoubin Yu, Taojiannan Yang, Lincoln Spencer, Yapeng Tian, Ajmal Saeed Mian, Mohit Bansal, Chen Chen
Abstract: In this paper, we introduce Motion-Grounded Video Reasoning, a new motion understanding task that requires generating visual answers (video segmentation masks) according to the input question, and hence needs implicit spatiotemporal reasoning and grounding. This task extends existing spatiotemporal grounding work focusing on explicit action/motion grounding, to a more general format by enabling implicit reasoning via questions. To facilitate the development of the new task, we collect a large-scale dataset called GROUNDMORE, which comprises 1,715 video clips, 249K object masks that are deliberately designed with 4 question types (Causal, Sequential, Counterfactual, and Descriptive) for benchmarking deep and comprehensive motion reasoning abilities. GROUNDMORE uniquely requires models to generate visual answers, providing a more concrete and visually interpretable response than plain texts. It evaluates models on both spatiotemporal grounding and reasoning, fostering to address complex challenges in motion-related video reasoning, temporal perception, and pixel-level understanding. Furthermore, we introduce a novel baseline model named Motion-Grounded Video Reasoning Assistant (MORA). MORA incorporates the multimodal reasoning ability from the Multimodal LLM, the pixel-level perception capability from the grounding model (SAM), and the temporal perception ability from a lightweight localization head. MORA achieves respectable performance on GROUNDMORE outperforming the best existing visual grounding baseline model by an average of 21.5% relatively. We hope this novel and challenging task will pave the way for future advancements in robust and general motion understanding via video reasoning segmentation
Authors: Yuanbin Man, Ying Huang, Chengming Zhang, Bingzhe Li, Wei Niu, Miao Yin
Abstract: The advancements in large language models (LLMs) have propelled the improvement of video understanding tasks by incorporating LLMs with visual models. However, most existing LLM-based models (e.g., VideoLLaMA, VideoChat) are constrained to processing short-duration videos. Recent attempts to understand long-term videos by extracting and compressing visual features into a fixed memory size. Nevertheless, those methods leverage only visual modality to merge video tokens and overlook the correlation between visual and textual queries, leading to difficulties in effectively handling complex question-answering tasks. To address the challenges of long videos and complex prompts, we propose AdaCM$^2$, which, for the first time, introduces an adaptive cross-modality memory reduction approach to video-text alignment in an auto-regressive manner on video streams. Our extensive experiments on various video understanding tasks, such as video captioning, video question answering, and video classification, demonstrate that AdaCM$^2$ achieves state-of-the-art performance across multiple datasets while significantly reducing memory usage. Notably, it achieves a 4.5% improvement across multiple tasks in the LVU dataset with a GPU memory consumption reduction of up to 65%.
Authors: Ruoyu Chen, Siyuan Liang, Jingzhi Li, Shiming Liu, Maosen Li, Zhen Huang, Hua Zhang, Xiaochun Cao
Abstract: Advances in multimodal pre-training have propelled object-level foundation models, such as Grounding DINO and Florence-2, in tasks like visual grounding and object detection. However, interpreting these models' decisions has grown increasingly challenging. Existing interpretable attribution methods for object-level task interpretation have notable limitations: (1) gradient-based methods lack precise localization due to visual-textual fusion in foundation models, and (2) perturbation-based methods produce noisy saliency maps, limiting fine-grained interpretability. To address these, we propose a Visual Precision Search method that generates accurate attribution maps with fewer regions. Our method bypasses internal model parameters to overcome attribution issues from multimodal fusion, dividing inputs into sparse sub-regions and using consistency and collaboration scores to accurately identify critical decision-making regions. We also conducted a theoretical analysis of the boundary guarantees and scope of applicability of our method. Experiments on RefCOCO, MS COCO, and LVIS show our approach enhances object-level task interpretability over SOTA for Grounding DINO and Florence-2 across various evaluation metrics, with faithfulness gains of 23.7%, 31.6%, and 20.1% on MS COCO, LVIS, and RefCOCO for Grounding DINO, and 102.9% and 66.9% on MS COCO and RefCOCO for Florence-2. Additionally, our method can interpret failures in visual grounding and object detection tasks, surpassing existing methods across multiple evaluation metrics. The code will be released at https://github.com/RuoyuChen10/VPS.
Authors: Yuncong Yang, Han Yang, Jiachen Zhou, Peihao Chen, Hongxin Zhang, Yilun Du, Chuang Gan
Abstract: Constructing compact and informative 3D scene representations is essential for effective embodied exploration and reasoning, especially in complex environments over extended periods. Existing representations, such as object-centric 3D scene graphs, oversimplify spatial relationships by modeling scenes as isolated objects with restrictive textual relationships, making it difficult to address queries requiring nuanced spatial understanding. Moreover, these representations lack natural mechanisms for active exploration and memory management, hindering their application to lifelong autonomy. In this work, we propose 3D-Mem, a novel 3D scene memory framework for embodied agents. 3D-Mem employs informative multi-view images, termed Memory Snapshots, to represent the scene and capture rich visual information of explored regions. It further integrates frontier-based exploration by introducing Frontier Snapshots-glimpses of unexplored areas-enabling agents to make informed decisions by considering both known and potential new information. To support lifelong memory in active exploration settings, we present an incremental construction pipeline for 3D-Mem, as well as a memory retrieval technique for memory management. Experimental results on three benchmarks demonstrate that 3D-Mem significantly enhances agents' exploration and reasoning capabilities in 3D environments, highlighting its potential for advancing applications in embodied AI.
Authors: Xiao An, Jiaxing Sun, Zihan Gui, Wei He
Abstract: The rapid advancement of Large Vision-Language Models (VLMs), both general-domain models and those specifically tailored for remote sensing, has demonstrated exceptional perception and reasoning capabilities in Earth observation tasks. However, a benchmark for systematically evaluating their capabilities in this domain is still lacking. To bridge this gap, we propose CHOICE, an extensive benchmark designed to objectively evaluate the hierarchical remote sensing capabilities of VLMs. Focusing on 2 primary capability dimensions essential to remote sensing: perception and reasoning, we further categorize 6 secondary dimensions and 23 leaf tasks to ensure a well-rounded assessment coverage. CHOICE guarantees the quality of all 10,507 problems through a rigorous process of data collection from 50 globally distributed cities, question construction and quality control. The newly curated data and the format of multiple-choice questions with definitive answers allow for an objective and straightforward performance assessment. Our evaluation of 3 proprietary and 21 open-source VLMs highlights their critical limitations within this specialized context. We hope that CHOICE will serve as a valuable resource and offer deeper insights into the challenges and potential of VLMs in the field of remote sensing. We will release CHOICE at https://github.com/ShawnAn-WHU/CHOICE.
Authors: Ana\"is Halin, S\'ebastien Pi\'erard, Anthony Cioppa, Marc Van Droogenbroeck
Abstract: Properly understanding the performances of classifiers is essential in various scenarios. However, the literature often relies only on one or two standard scores to compare classifiers, which fails to capture the nuances of application-specific requirements. The Tile is a recently introduced visualization tool organizing an infinity of ranking scores into a 2D map. Thanks to the Tile, it is now possible to compare classifiers efficiently, displaying all possible application-specific preferences instead of having to rely on a pair of scores. This hitchhiker's guide to understanding the performances of two-class classifiers presents four scenarios showcasing different user profiles: a theoretical analyst, a method designer, a benchmarker, and an application developer. We introduce several interpretative flavors adapted to the user's needs by mapping different values on the Tile. We illustrate this guide by ranking and analyzing the performances of 74 state-of-the-art semantic segmentation models through the perspective of the four scenarios. Through these user profiles, we demonstrate that the Tile effectively captures the behavior of classifiers in a single visualization, while accommodating an infinite number of ranking scores. Code for mapping the different Tile flavors is available in supplementary material.
Authors: Chun-Peng Chang, Alain Pagani, Didier Stricker
Abstract: Multimodal Large Language Models (MLLMs) have made significant progress in tasks such as image captioning and question answering. However, while these models can generate realistic captions, they often struggle with providing precise instructions, particularly when it comes to localizing and disambiguating objects in complex 3D environments. This capability is critical as MLLMs become more integrated with collaborative robotic systems. In scenarios where a target object is surrounded by similar objects (distractors), robots must deliver clear, spatially-aware instructions to guide humans effectively. We refer to this challenge as contextual object localization and disambiguation, which imposes stricter constraints than conventional 3D dense captioning, especially regarding ensuring target exclusivity. In response, we propose simple yet effective techniques to enhance the model's ability to localize and disambiguate target objects. Our approach not only achieves state-of-the-art performance on conventional metrics that evaluate sentence similarity, but also demonstrates improved 3D spatial understanding through 3D visual grounding model.
Authors: M. Hamza Mughal, Rishabh Dabral, Merel C. J. Scholman, Vera Demberg, Christian Theobalt
Abstract: Non-verbal communication often comprises of semantically rich gestures that help convey the meaning of an utterance. Producing such semantic co-speech gestures has been a major challenge for the existing neural systems that can generate rhythmic beat gestures, but struggle to produce semantically meaningful gestures. Therefore, we present RAG-Gesture, a diffusion-based gesture generation approach that leverages Retrieval Augmented Generation (RAG) to produce natural-looking and semantically rich gestures. Our neuro-explicit gesture generation approach is designed to produce semantic gestures grounded in interpretable linguistic knowledge. We achieve this by using explicit domain knowledge to retrieve exemplar motions from a database of co-speech gestures. Once retrieved, we then inject these semantic exemplar gestures into our diffusion-based gesture generation pipeline using DDIM inversion and retrieval guidance at the inference time without any need of training. Further, we propose a control paradigm for guidance, that allows the users to modulate the amount of influence each retrieval insertion has over the generated sequence. Our comparative evaluations demonstrate the validity of our approach against recent gesture generation approaches. The reader is urged to explore the results on our project page.
Authors: Mehdi Noroozi, Isma Hadji, Victor Escorcia, Anestis Zaganidis, Brais Martinez, Georgios Tzimiropoulos
Abstract: There has been immense progress recently in the visual quality of Stable Diffusion-based Super Resolution (SD-SR). However, deploying large diffusion models on computationally restricted devices such as mobile phones remains impractical due to the large model size and high latency. This is compounded for SR as it often operates at high res (e.g. 4Kx3K). In this work, we introduce Edge-SD-SR, the first parameter efficient and low latency diffusion model for image super-resolution. Edge-SD-SR consists of ~169M parameters, including UNet, encoder and decoder, and has a complexity of only ~142 GFLOPs. To maintain a high visual quality on such low compute budget, we introduce a number of training strategies: (i) A novel conditioning mechanism on the low resolution input, coined bidirectional conditioning, which tailors the SD model for the SR task. (ii) Joint training of the UNet and encoder, while decoupling the encodings of the HR and LR images and using a dedicated schedule. (iii) Finetuning the decoder using the UNet's output to directly tailor the decoder to the latents obtained at inference time. Edge-SD-SR runs efficiently on device, e.g. it can upscale a 128x128 patch to 512x512 in 38 msec while running on a Samsung S24 DSP, and of a 512x512 to 2048x2048 (requiring 25 model evaluations) in just ~1.1 sec. Furthermore, we show that Edge-SD-SR matches or even outperforms state-of-the-art SR approaches on the most established SR benchmarks.
Authors: Yanzhe Lyu, Kai Cheng, Xin Kang, Xuejin Chen
Abstract: Recently, 3D Gaussian Splatting (3D-GS) has prevailed in novel view synthesis, achieving high fidelity and efficiency. However, it often struggles to capture rich details and complete geometry. Our analysis reveals that the 3D-GS densification operation lacks adaptiveness and faces a dilemma between geometry coverage and detail recovery. To address this, we introduce a novel densification operation, residual split, which adds a downscaled Gaussian as a residual. Our approach is capable of adaptively retrieving details and complementing missing geometry. To further support this method, we propose a pipeline named ResGS. Specifically, we integrate a Gaussian image pyramid for progressive supervision and implement a selection scheme that prioritizes the densification of coarse Gaussians over time. Extensive experiments demonstrate that our method achieves SOTA rendering quality. Consistent performance improvements can be achieved by applying our residual split on various 3D-GS variants, underscoring its versatility and potential for broader application in 3D-GS-based applications.
Authors: Weixiang Zhang, Shuzhao Xie, Chengwei Ren, Siyi Xie, Chen Tang, Shijia Ge, Mingzi Wang, Zhi Wang
Abstract: We propose EVOlutionary Selector (EVOS), an efficient training paradigm for accelerating Implicit Neural Representation (INR). Unlike conventional INR training that feeds all samples through the neural network in each iteration, our approach restricts training to strategically selected points, reducing computational overhead by eliminating redundant forward passes. Specifically, we treat each sample as an individual in an evolutionary process, where only those fittest ones survive and merit inclusion in training, adaptively evolving with the neural network dynamics. While this is conceptually similar to Evolutionary Algorithms, their distinct objectives (selection for acceleration vs. iterative solution optimization) require a fundamental redefinition of evolutionary mechanisms for our context. In response, we design sparse fitness evaluation, frequency-guided crossover, and augmented unbiased mutation to comprise EVOS. These components respectively guide sample selection with reduced computational cost, enhance performance through frequency-domain balance, and mitigate selection bias from cached evaluation. Extensive experiments demonstrate that our method achieves approximately 48%-66% reduction in training time while ensuring superior convergence without additional cost, establishing state-of-the-art acceleration among recent sampling-based strategies.
Authors: Rick Akkerman, Haiwen Feng, Michael J. Black, Dimitrios Tzionas, Victoria Fern\'andez Abrevaya
Abstract: Predicting the dynamics of interacting objects is essential for both humans and intelligent systems. However, existing approaches are limited to simplified, toy settings and lack generalizability to complex, real-world environments. Recent advances in generative models have enabled the prediction of state transitions based on interventions, but focus on generating a single future state which neglects the continuous dynamics resulting from the interaction. To address this gap, we propose InterDyn, a novel framework that generates videos of interactive dynamics given an initial frame and a control signal encoding the motion of a driving object or actor. Our key insight is that large video generation models can act as both neural renderers and implicit physics ``simulators'', having learned interactive dynamics from large-scale video data. To effectively harness this capability, we introduce an interactive control mechanism that conditions the video generation process on the motion of the driving entity. Qualitative results demonstrate that InterDyn generates plausible, temporally consistent videos of complex object interactions while generalizing to unseen objects. Quantitative evaluations show that InterDyn outperforms baselines that focus on static state transitions. This work highlights the potential of leveraging video generative models as implicit physics engines. Project page: https://interdyn.is.tue.mpg.de/
Authors: Austin Stone, Hagen Soltau, Robert Geirhos, Xi Yi, Ye Xia, Bingyi Cao, Kaifeng Chen, Abhijit Ogale, Jonathon Shlens
Abstract: Visual imagery does not consist of solitary objects, but instead reflects the composition of a multitude of fluid concepts. While there have been great advances in visual representation learning, such advances have focused on building better representations for a small number of discrete objects bereft of an understanding of how these objects are interacting. One can observe this limitation in representations learned through captions or contrastive learning -- where the learned model treats an image essentially as a bag of words. Several works have attempted to address this limitation through the development of bespoke learned architectures to directly address the shortcomings in compositional learning. In this work, we focus on simple, and scalable approaches. In particular, we demonstrate that by substantially improving weakly labeled data, i.e. captions, we can vastly improve the performance of standard contrastive learning approaches. Previous CLIP models achieved near chance rate on challenging tasks probing compositional learning. However, our simple approach boosts performance of CLIP substantially and surpasses all bespoke architectures. Furthermore, we showcase our results on a relatively new captioning benchmark derived from DOCCI. We demonstrate through a series of ablations that a standard CLIP model trained with enhanced data may demonstrate impressive performance on image retrieval tasks.
Authors: Xiang Fang, Wanlong Fang, Changshuo Wang, Daizong Liu, Keke Tang, Jianfeng Dong, Pan Zhou, Beibei Li
Abstract: Given some video-query pairs with untrimmed videos and sentence queries, temporal sentence grounding (TSG) aims to locate query-relevant segments in these videos. Although previous respectable TSG methods have achieved remarkable success, they train each video-query pair separately and ignore the relationship between different pairs. We observe that the similar video/query content not only helps the TSG model better understand and generalize the cross-modal representation but also assists the model in locating some complex video-query pairs. Previous methods follow a single-thread framework that cannot co-train different pairs and usually spends much time re-obtaining redundant knowledge, limiting their real-world applications. To this end, in this paper, we pose a brand-new setting: Multi-Pair TSG, which aims to co-train these pairs. In particular, we propose a novel video-query co-training approach, Multi-Thread Knowledge Transfer Network, to locate a variety of video-query pairs effectively and efficiently. Firstly, we mine the spatial and temporal semantics across different queries to cooperate with each other. To learn intra- and inter-modal representations simultaneously, we design a cross-modal contrast module to explore the semantic consistency by a self-supervised strategy. To fully align visual and textual representations between different pairs, we design a prototype alignment strategy to 1) match object prototypes and phrase prototypes for spatial alignment, and 2) align activity prototypes and sentence prototypes for temporal alignment. Finally, we develop an adaptive negative selection module to adaptively generate a threshold for cross-modal matching. Extensive experiments show the effectiveness and efficiency of our proposed method.
Authors: Tianyun Zhong, Chao Liang, Jianwen Jiang, Gaojie Lin, Jiaqi Yang, Zhou Zhao
Abstract: Diffusion-based audio-driven talking avatar methods have recently gained attention for their high-fidelity, vivid, and expressive results. However, their slow inference speed limits practical applications. Despite the development of various distillation techniques for diffusion models, we found that naive diffusion distillation methods do not yield satisfactory results. Distilled models exhibit reduced robustness with open-set input images and a decreased correlation between audio and video compared to teacher models, undermining the advantages of diffusion models. To address this, we propose FADA (Fast Diffusion Avatar Synthesis with Mixed-Supervised Multi-CFG Distillation). We first designed a mixed-supervised loss to leverage data of varying quality and enhance the overall model capability as well as robustness. Additionally, we propose a multi-CFG distillation with learnable tokens to utilize the correlation between audio and reference image conditions, reducing the threefold inference runs caused by multi-CFG with acceptable quality degradation. Extensive experiments across multiple datasets show that FADA generates vivid videos comparable to recent diffusion model-based methods while achieving an NFE speedup of 4.17-12.5 times. Demos are available at our webpage http://fadavatar.github.io.
Authors: Kiet A. Nguyen, Adheesh Juvekar, Tianjiao Yu, Muntasir Wahed, Ismini Lourentzou
Abstract: Recent advances in Large Vision-Language Models (LVLMs) have enabled general-purpose vision tasks through visual instruction tuning. While existing LVLMs can generate segmentation masks from text prompts for single images, they struggle with segmentation-grounded reasoning across images, especially at finer granularities such as object parts. In this paper, we introduce the new task of part-focused semantic co-segmentation, which involves identifying and segmenting common objects, as well as common and unique object parts across images. To address this task, we present CALICO, the first LVLM designed for multi-image part-level reasoning segmentation. CALICO features two key components, a novel Correspondence Extraction Module that identifies semantic part-level correspondences, and Correspondence Adaptation Modules that embed this information into the LVLM to facilitate multi-image understanding in a parameter-efficient manner. To support training and evaluation, we curate MixedParts, a large-scale multi-image segmentation dataset containing $\sim$2.4M samples across $\sim$44K images spanning diverse object and part categories. Experimental results demonstrate that CALICO, with just 0.3% of its parameters finetuned, achieves strong performance on this challenging task.
Authors: Tomer Garber, Tom Tirer
Abstract: In recent years, it has become popular to tackle image restoration tasks with a single pretrained diffusion model (DM) and data-fidelity guidance, instead of training a dedicated deep neural network per task. However, such "zero-shot" restoration schemes currently require many Neural Function Evaluations (NFEs) for performing well, which may be attributed to the many NFEs needed in the original generative functionality of the DMs. Recently, faster variants of DMs have been explored for image generation. These include Consistency Models (CMs), which can generate samples via a couple of NFEs. However, existing works that use guided CMs for restoration still require tens of NFEs or fine-tuning of the model per task that leads to performance drop if the assumptions during the fine-tuning are not accurate. In this paper, we propose a zero-shot restoration scheme that uses CMs and operates well with as little as 4 NFEs. It is based on a wise combination of several ingredients: better initialization, back-projection guidance, and above all a novel noise injection mechanism. We demonstrate the advantages of our approach for image super-resolution, deblurring and inpainting. Interestingly, we show that the usefulness of our noise injection technique goes beyond CMs: it can also mitigate the performance degradation of existing guided DM methods when reducing their NFE count.
Authors: Lingzhi Yuan, Xiaojun Jia, Yihao Huang, Wei Dong, Yang Liu
Abstract: Text-to-image (T2I) models have been shown to be vulnerable to misuse, particularly in generating not-safe-for-work (NSFW) content, raising serious ethical concerns. In this work, we present PromptGuard, a novel content moderation technique that draws inspiration from the system prompt mechanism in large language models (LLMs) for safety alignment. Unlike LLMs, T2I models lack a direct interface for enforcing behavioral guidelines. Our key idea is to optimize a safety soft prompt that functions as an implicit system prompt within the T2I model's textual embedding space. This universal soft prompt (P*) directly moderates NSFW inputs, enabling safe yet realistic image generation without altering the inference efficiency or requiring proxy models. Extensive experiments across three datasets demonstrate that PromptGuard effectively mitigates NSFW content generation while preserving high-quality benign outputs. PromptGuard achieves 7.8 times faster than prior content moderation methods, surpassing eight state-of-the-art defenses with an optimal unsafe ratio down to 5.84%.
Authors: Mert Keser, Halil Ibrahim Orhan, Niki Amini-Naieni, Gesina Schwalbe, Alois Knoll, Matthias Rottmann
Abstract: Deep neural networks (DNNs) remain challenged by distribution shifts in complex open-world domains like automated driving (AD): Robustness against yet unknown novel objects (semantic shift) or styles like lighting conditions (covariate shift) cannot be guaranteed. Hence, reliable operation-time monitors for identification of out-of-training-data-distribution (OOD) scenarios are imperative. Current approaches for OOD classification are untested for complex domains like AD, are limited in the kinds of shifts they detect, or even require supervision with OOD samples. To prepare for unanticipated shifts, we instead establish a framework around a principled, unsupervised and model-agnostic method that unifies detection of semantic and covariate shifts: Find a full model of the training data's feature distribution, to then use its density at new points as in-distribution (ID) score. To implement this, we propose to combine Vision Foundation Models (VFMs) as feature extractors with density modeling techniques. Through a comprehensive benchmark of 4 VFMs with different backbone architectures and 5 density-modeling techniques against established baselines, we provide the first systematic evaluation of OOD classification capabilities of VFMs across diverse conditions. A comparison with state-of-the-art binary OOD classification methods reveals that VFM embeddings with density estimation outperform existing approaches in identifying OOD inputs. Additionally, we show that our method detects high-risk inputs likely to cause errors in downstream tasks, thereby improving overall performance. Overall, VFMs, when coupled with robust density modeling techniques, are promising to realize model-agnostic, unsupervised, reliable safety monitors in complex vision tasks
Authors: Bowen Wen, Matthew Trepte, Joseph Aribido, Jan Kautz, Orazio Gallo, Stan Birchfield
Abstract: Tremendous progress has been made in deep stereo matching to excel on benchmark datasets through per-domain fine-tuning. However, achieving strong zero-shot generalization - a hallmark of foundation models in other computer vision tasks - remains challenging for stereo matching. We introduce FoundationStereo, a foundation model for stereo depth estimation designed to achieve strong zero-shot generalization. To this end, we first construct a large-scale (1M stereo pairs) synthetic training dataset featuring large diversity and high photorealism, followed by an automatic self-curation pipeline to remove ambiguous samples. We then design a number of network architecture components to enhance scalability, including a side-tuning feature backbone that adapts rich monocular priors from vision foundation models to mitigate the sim-to-real gap, and long-range context reasoning for effective cost volume filtering. Together, these components lead to strong robustness and accuracy across domains, establishing a new standard in zero-shot stereo depth estimation. Project page: https://nvlabs.github.io/FoundationStereo/
Authors: Ludovica Schaerf, Andrea Alfarano, Fabrizio Silvestri, Leonardo Impett
Abstract: Recent advances in diffusion models for image generation have led to detailed examinations of several components within the U-Net architecture for image editing. While previous studies have focused on the bottleneck layer (h-space), cross-attention, self-attention, and decoding layers, the overall role of the skip connections of the U-Net itself has not been specifically addressed. We conduct thorough analyses on the role of the skip connections and find that the residual connections passed by the third encoder block carry most of the spatial information of the reconstructed image, splitting the content from the style, passed by the remaining stream in the opposed decoding layer. We show that injecting the representations from this block can be used for text-based editing, precise modifications, and style transfer. We compare our method, SkipInject, to state-of-the-art style transfer and image editing methods and demonstrate that our method obtains the best content alignment and optimal structural preservation tradeoff.
Authors: Youming Deng, Wenqi Xian, Guandao Yang, Leonidas Guibas, Gordon Wetzstein, Steve Marschner, Paul Debevec
Abstract: In this paper, we present a self-calibrating framework that jointly optimizes camera parameters, lens distortion and 3D Gaussian representations, enabling accurate and efficient scene reconstruction. In particular, our technique enables high-quality scene reconstruction from Large field-of-view (FOV) imagery taken with wide-angle lenses, allowing the scene to be modeled from a smaller number of images. Our approach introduces a novel method for modeling complex lens distortions using a hybrid network that combines invertible residual networks with explicit grids. This design effectively regularizes the optimization process, achieving greater accuracy than conventional camera models. Additionally, we propose a cubemap-based resampling strategy to support large FOV images without sacrificing resolution or introducing distortion artifacts. Our method is compatible with the fast rasterization of Gaussian Splatting, adaptable to a wide variety of camera lens distortion, and demonstrates state-of-the-art performance on both synthetic and real-world datasets.
Authors: Yisheng He, Xiaodong Gu, Xiaodan Ye, Chao Xu, Zhengyi Zhao, Yuan Dong, Weihao Yuan, Zilong Dong, Liefeng Bo
Abstract: We present LAM, an innovative Large Avatar Model for animatable Gaussian head reconstruction from a single image. Unlike previous methods that require extensive training on captured video sequences or rely on auxiliary neural networks for animation and rendering during inference, our approach generates Gaussian heads that are immediately animatable and renderable. Specifically, LAM creates an animatable Gaussian head in a single forward pass, enabling reenactment and rendering without additional networks or post-processing steps. This capability allows for seamless integration into existing rendering pipelines, ensuring real-time animation and rendering across a wide range of platforms, including mobile phones. The centerpiece of our framework is the canonical Gaussian attributes generator, which utilizes FLAME canonical points as queries. These points interact with multi-scale image features through a Transformer to accurately predict Gaussian attributes in the canonical space. The reconstructed canonical Gaussian avatar can then be animated utilizing standard linear blend skinning (LBS) with corrective blendshapes as the FLAME model did and rendered in real-time on various platforms. Our experimental results demonstrate that LAM outperforms state-of-the-art methods on existing benchmarks. Our code and video are available at https://aigc3d.github.io/projects/LAM/
Authors: Aayush Dhakal, Srikumar Sastry, Subash Khanal, Adeel Ahmad, Eric Xing, Nathan Jacobs
Abstract: The choice of representation for geographic location significantly impacts the accuracy of models for a broad range of geospatial tasks, including fine-grained species classification, population density estimation, and biome classification. Recent works like SatCLIP and GeoCLIP learn such representations by contrastively aligning geolocation with co-located images. While these methods work exceptionally well, in this paper, we posit that the current training strategies fail to fully capture the important visual features. We provide an information-theoretic perspective on why the resulting embeddings from these methods discard crucial visual information that is important for many downstream tasks. To solve this problem, we propose a novel retrieval-augmented strategy called RANGE. We build our method on the intuition that the visual features of a location can be estimated by combining the visual features from multiple similar-looking locations. We evaluate our method across a wide variety of tasks. Our results show that RANGE outperforms the existing state-of-the-art models with significant margins in most tasks. We show gains of up to 13.1% on classification tasks and 0.145 $R^2$ on regression tasks. All our code and models will be made available at: https://github.com/mvrl/RANGE.
Authors: David T. Hoffmann, Syed Haseeb Raza, Hanqiu Jiang, Denis Tananaev, Steffen Klingenhoefer, Martin Meinke
Abstract: Scene flow estimation is a foundational task for many robotic applications, including robust dynamic object detection, automatic labeling, and sensor synchronization. Two types of approaches to the problem have evolved: 1) Supervised and 2) optimization-based methods. Supervised methods are fast during inference and achieve high-quality results, however, they are limited by the need for large amounts of labeled training data and are susceptible to domain gaps. In contrast, unsupervised test-time optimization methods do not face the problem of domain gaps but usually suffer from substantial runtime, exhibit artifacts, or fail to converge to the right solution. In this work, we mitigate several limitations of existing optimization-based methods. To this end, we 1) introduce a simple voxel grid-based model that improves over the standard MLP-based formulation in multiple dimensions and 2) introduce a new multiframe loss formulation. 3) We combine both contributions in our new method, termed Floxels. On the Argoverse 2 benchmark, Floxels is surpassed only by EulerFlow among unsupervised methods while achieving comparable performance at a fraction of the computational cost. Floxels achieves a massive speedup of more than ~60 - 140x over EulerFlow, reducing the runtime from a day to 10 minutes per sequence. Over the faster but low-quality baseline, NSFP, Floxels achieves a speedup of ~14x.
Authors: Peizhen Zheng, Longfei Wei, Dongjing Jiang, Jianfei Zhang
Abstract: The accurate reconstruction of dynamic street scenes is critical for applications in autonomous driving, augmented reality, and virtual reality. Traditional methods relying on dense point clouds and triangular meshes struggle with moving objects, occlusions, and real-time processing constraints, limiting their effectiveness in complex urban environments. While multi-view stereo and neural radiance fields have advanced 3D reconstruction, they face challenges in computational efficiency and handling scene dynamics. This paper proposes a novel 3D Gaussian point distribution method for dynamic street scene reconstruction. Our approach introduces an adaptive transparency mechanism that eliminates moving objects while preserving high-fidelity static scene details. Additionally, iterative refinement of Gaussian point distribution enhances geometric accuracy and texture representation. We integrate directional encoding with spatial position optimization to optimize storage and rendering efficiency, reducing redundancy while maintaining scene integrity. Experimental results demonstrate that our method achieves high reconstruction quality, improved rendering performance, and adaptability in large-scale dynamic environments. These contributions establish a robust framework for real-time, high-precision 3D reconstruction, advancing the practicality of dynamic scene modeling across multiple applications.
Authors: Guangqian Guo, Yong Guo, Xuehui Yu, Wenbo Li, Yaoxing Wang, Shan Gao
Abstract: Despite their success, Segment Anything Models (SAMs) experience significant performance drops on severely degraded, low-quality images, limiting their effectiveness in real-world scenarios. To address this, we propose GleSAM, which utilizes Generative Latent space Enhancement to boost robustness on low-quality images, thus enabling generalization across various image qualities. Specifically, we adapt the concept of latent diffusion to SAM-based segmentation frameworks and perform the generative diffusion process in the latent space of SAM to reconstruct high-quality representation, thereby improving segmentation. Additionally, we introduce two techniques to improve compatibility between the pre-trained diffusion model and the segmentation framework. Our method can be applied to pre-trained SAM and SAM2 with only minimal additional learnable parameters, allowing for efficient optimization. We also construct the LQSeg dataset with a greater diversity of degradation types and levels for training and evaluating the model. Extensive experiments demonstrate that GleSAM significantly improves segmentation robustness on complex degradations while maintaining generalization to clear images. Furthermore, GleSAM also performs well on unseen degradations, underscoring the versatility of our approach and dataset.
Authors: Yanis Benidir, Nicolas Gonthier, Clement Mallet
Abstract: Bi-temporal change detection at scale based on Very High Resolution (VHR) images is crucial for Earth monitoring. This remains poorly addressed so far: methods either require large volumes of annotated data (semantic case), or are limited to restricted datasets (binary set-ups). Most approaches do not exhibit the versatility required for temporal and spatial adaptation: simplicity in architecture design and pretraining on realistic and comprehensive datasets. Synthetic datasets are the key solution but still fail to handle complex and diverse scenes. In this paper, we present HySCDG a generative pipeline for creating a large hybrid semantic change detection dataset that contains both real VHR images and inpainted ones, along with land cover semantic map at both dates and the change map. Being semantically and spatially guided, HySCDG generates realistic images, leading to a comprehensive and hybrid transfer-proof dataset FSC-180k. We evaluate FSC-180k on five change detection cases (binary and semantic), from zero-shot to mixed and sequential training, and also under low data regime training. Experiments demonstrate that pretraining on our hybrid dataset leads to a significant performance boost, outperforming SyntheWorld, a fully synthetic dataset, in every configuration. All codes, models, and data are available here: https://yb23.github.io/projects/cywd/
Authors: Rong Wang, Fabian Prada, Ziyan Wang, Zhongshi Jiang, Chengxiang Yin, Junxuan Li, Shunsuke Saito, Igor Santesteban, Javier Romero, Rohan Joshi, Hongdong Li, Jason Saragih, Yaser Sheikh
Abstract: We present a novel method for reconstructing personalized 3D human avatars with realistic animation from only a few images. Due to the large variations in body shapes, poses, and cloth types, existing methods mostly require hours of per-subject optimization during inference, which limits their practical applications. In contrast, we learn a universal prior from over a thousand clothed humans to achieve instant feedforward generation and zero-shot generalization. Specifically, instead of rigging the avatar with shared skinning weights, we jointly infer personalized avatar shape, skinning weights, and pose-dependent deformations, which effectively improves overall geometric fidelity and reduces deformation artifacts. Moreover, to normalize pose variations and resolve coupled ambiguity between canonical shapes and skinning weights, we design a 3D canonicalization process to produce pixel-aligned initial conditions, which helps to reconstruct fine-grained geometric details. We then propose a multi-frame feature aggregation to robustly reduce artifacts introduced in canonicalization and fuse a plausible avatar preserving person-specific identities. Finally, we train the model in an end-to-end framework on a large-scale capture dataset, which contains diverse human subjects paired with high-quality 3D scans. Extensive experiments show that our method generates more authentic reconstruction and animation than state-of-the-arts, and can be directly generalized to inputs from casually taken phone photos. Project page and code is available at https://github.com/rongakowang/FRESA.
Authors: Amaya Gallagher-Syed, Henry Senior, Omnia Alwazzan, Elena Pontarini, Michele Bombardieri, Costantino Pitzalis, Myles J. Lewis, Michael R. Barnes, Luca Rossi, Gregory Slabaugh
Abstract: The development of biologically interpretable and explainable models remains a key challenge in computational pathology, particularly for multistain immunohistochemistry (IHC) analysis. We present BioX-CPath, an explainable graph neural network architecture for whole slide image (WSI) classification that leverages both spatial and semantic features across multiple stains. At its core, BioX-CPath introduces a novel Stain-Aware Attention Pooling (SAAP) module that generates biologically meaningful, stain-aware patient embeddings. Our approach achieves state-of-the-art performance on both Rheumatoid Arthritis and Sjogren's Disease multistain datasets. Beyond performance metrics, BioX-CPath provides interpretable insights through stain attention scores, entropy measures, and stain interaction scores, that permit measuring model alignment with known pathological mechanisms. This biological grounding, combined with strong classification performance, makes BioX-CPath particularly suitable for clinical applications where interpretability is key. Source code and documentation can be found at: https://github.com/AmayaGS/BioX-CPath.
Authors: Edwin Tay, Nazli T\"umer, Amir A. Zadpoor
Abstract: Living biological tissue is a complex system, constantly growing and changing in response to external and internal stimuli. These processes lead to remarkable and intricate changes in shape. Modeling and understanding both natural and pathological (or abnormal) changes in the shape of anatomical structures is highly relevant, with applications in diagnostic, prognostic, and therapeutic healthcare. Nevertheless, modeling the longitudinal shape change of biological tissue is a non-trivial task due to its inherent nonlinear nature. In this review, we highlight several existing methodologies and tools for modeling longitudinal shape change (i.e., spatiotemporal shape modeling). These methods range from diffeomorphic metric mapping to deep-learning based approaches (e.g., autoencoders, generative networks, recurrent neural networks, etc.). We discuss the synergistic combinations of existing technologies and potential directions for future research, underscoring key deficiencies in the current research landscape.
Authors: Boyi Ma, Yanguang Zhao, Jie Wang, Guankun Wang, Kun Yuan, Tong Chen, Long Bai, Hongliang Ren
Abstract: The DeepSeek models have shown exceptional performance in general scene understanding, question-answering (QA), and text generation tasks, owing to their efficient training paradigm and strong reasoning capabilities. In this study, we investigate the dialogue capabilities of the DeepSeek model in robotic surgery scenarios, focusing on tasks such as Single Phrase QA, Visual QA, and Detailed Description. The Single Phrase QA tasks further include sub-tasks such as surgical instrument recognition, action understanding, and spatial position analysis. We conduct extensive evaluations using publicly available datasets, including EndoVis18 and CholecT50, along with their corresponding dialogue data. Our empirical study shows that, compared to existing general-purpose multimodal large language models, DeepSeek-VL2 performs better on complex understanding tasks in surgical scenes. Additionally, although DeepSeek-V3 is purely a language model, we find that when image tokens are directly inputted, the model demonstrates better performance on single-sentence QA tasks. However, overall, the DeepSeek models still fall short of meeting the clinical requirements for understanding surgical scenes. Under general prompts, DeepSeek models lack the ability to effectively analyze global surgical concepts and fail to provide detailed insights into surgical scenarios. Based on our observations, we argue that the DeepSeek models are not ready for vision-language tasks in surgical contexts without fine-tuning on surgery-specific datasets.
Authors: Shota Hirose, Kazuki Kotoyori, Kasidis Arunruangsirilert, Fangzheng Lin, Heming Sun, Jiro Katto
Abstract: Transmission latency significantly affects users' quality of experience in real-time interaction and actuation. As latency is principally inevitable, video prediction can be utilized to mitigate the latency and ultimately enable zero-latency transmission. However, most of the existing video prediction methods are computationally expensive and impractical for real-time applications. In this work, we therefore propose real-time video prediction towards the zero-latency interaction over networks, called IFRVP (Intermediate Feature Refinement Video Prediction). Firstly, we propose three training methods for video prediction that extend frame interpolation models, where we utilize a simple convolution-only frame interpolation network based on IFRNet. Secondly, we introduce ELAN-based residual blocks into the prediction models to improve both inference speed and accuracy. Our evaluations show that our proposed models perform efficiently and achieve the best trade-off between prediction accuracy and computational speed among the existing video prediction methods. A demonstration movie is also provided at http://bit.ly/IFRVPDemo. The code will be released at https://github.com/FykAikawa/IFRVP.
URLs: http://bit.ly/IFRVPDemo., https://github.com/FykAikawa/IFRVP.
Authors: Xindi Yang, Baolu Li, Yiming Zhang, Zhenfei Yin, Lei Bai, Liqian Ma, Zhiyong Wang, Jianfei Cai, Tien-Tsin Wong, Huchuan Lu, Xu Jia
Abstract: Video diffusion models (VDMs) have advanced significantly in recent years, enabling the generation of highly realistic videos and drawing the attention of the community in their potential as world simulators. However, despite their capabilities, VDMs often fail to produce physically plausible videos due to an inherent lack of understanding of physics, resulting in incorrect dynamics and event sequences. To address this limitation, we propose a novel two-stage image-to-video generation framework that explicitly incorporates physics with vision and language informed physical prior. In the first stage, we employ a Vision Language Model (VLM) as a coarse-grained motion planner, integrating chain-of-thought and physics-aware reasoning to predict a rough motion trajectories/changes that approximate real-world physical dynamics while ensuring the inter-frame consistency. In the second stage, we use the predicted motion trajectories/changes to guide the video generation of a VDM. As the predicted motion trajectories/changes are rough, noise is added during inference to provide freedom to the VDM in generating motion with more fine details. Extensive experimental results demonstrate that our framework can produce physically plausible motion, and comparative evaluations highlight the notable superiority of our approach over existing methods. More video results are available on our Project Page: https://madaoer.github.io/projects/physically_plausible_video_generation.
URLs: https://madaoer.github.io/projects/physically_plausible_video_generation.
Authors: Xiaole Xian, Zhichao Liao, Qingyu Li, Wenyu Qin, Pengfei Wan, Weicheng Xie, Long Zeng, Linlin Shen, Pingfa Feng
Abstract: Fine-tuning a pre-trained Text-to-Image (T2I) model on a tailored portrait dataset is the mainstream method for text-driven customization of portrait attributes. Due to Semantic Pollution during fine-tuning, existing methods struggle to maintain the original model's behavior and achieve incremental learning while customizing target attributes. To address this issue, we propose SPF-Portrait, a pioneering work to purely understand customized semantics while eliminating semantic pollution in text-driven portrait customization. In our SPF-Portrait, we propose a dual-path pipeline that introduces the original model as a reference for the conventional fine-tuning path. Through contrastive learning, we ensure adaptation to target attributes and purposefully align other unrelated attributes with the original portrait. We introduce a novel Semantic-Aware Fine Control Map, which represents the precise response regions of the target semantics, to spatially guide the alignment process between the contrastive paths. This alignment process not only effectively preserves the performance of the original model but also avoids over-alignment. Furthermore, we propose a novel response enhancement mechanism to reinforce the performance of target attributes, while mitigating representation discrepancy inherent in direct cross-modal supervision. Extensive experiments demonstrate that SPF-Portrait achieves state-of-the-art performance. Project webpage: https://spf-portrait.github.io/SPF-Portrait/
Authors: Lukas Boehm, Jonas Leo Mueller, Christoffer Loeffler, Leo Schwinn, Bjoern Eskofier, Dario Zanca
Abstract: Understanding the perceptual invariances of artificial neural networks is essential for improving explainability and aligning models with human vision. Metamers - stimuli that are physically distinct yet produce identical neural activations - serve as a valuable tool for investigating these invariances. We introduce a novel approach to metamer generation by leveraging ensembles of artificial neural networks, capturing shared representational subspaces across diverse architectures, including convolutional neural networks and vision transformers. To characterize the properties of the generated metamers, we employ a suite of image-based metrics that assess factors such as semantic fidelity and naturalness. Our findings show that convolutional neural networks generate more recognizable and human-like metamers, while vision transformers produce realistic but less transferable metamers, highlighting the impact of architectural biases on representational invariances.
Authors: Fa-Ting Hong, Zunnan Xu, Zixiang Zhou, Jun Zhou, Xiu Li, Qin Lin, Qinglin Lu, Dan Xu
Abstract: Talking head synthesis is vital for virtual avatars and human-computer interaction. However, most existing methods are typically limited to accepting control from a single primary modality, restricting their practical utility. To this end, we introduce \textbf{ACTalker}, an end-to-end video diffusion framework that supports both multi-signals control and single-signal control for talking head video generation. For multiple control, we design a parallel mamba structure with multiple branches, each utilizing a separate driving signal to control specific facial regions. A gate mechanism is applied across all branches, providing flexible control over video generation. To ensure natural coordination of the controlled video both temporally and spatially, we employ the mamba structure, which enables driving signals to manipulate feature tokens across both dimensions in each branch. Additionally, we introduce a mask-drop strategy that allows each driving signal to independently control its corresponding facial region within the mamba structure, preventing control conflicts. Experimental results demonstrate that our method produces natural-looking facial videos driven by diverse signals and that the mamba layer seamlessly integrates multiple driving modalities without conflict. The project website can be found at \href{https://harlanhong.github.io/publications/actalker/index.html}{HERE}.
URLs: https://harlanhong.github.io/publications/actalker/index.html
Authors: Tiago Novello, Diana Aldana, Andre Araujo, Luiz Velho
Abstract: Sinusoidal neural networks have been shown effective as implicit neural representations (INRs) of low-dimensional signals, due to their smoothness and high representation capacity. However, initializing and training them remain empirical tasks which lack on deeper understanding to guide the learning process. To fill this gap, our work introduces a theoretical framework that explains the capacity property of sinusoidal networks and offers robust control mechanisms for initialization and training. Our analysis is based on a novel amplitude-phase expansion of the sinusoidal multilayer perceptron, showing how its layer compositions produce a large number of new frequencies expressed as integer combinations of the input frequencies. This relationship can be directly used to initialize the input neurons, as a form of spectral sampling, and to bound the network's spectrum while training. Our method, referred to as TUNER (TUNing sinusoidal nEtwoRks), greatly improves the stability and convergence of sinusoidal INR training, leading to detailed reconstructions, while preventing overfitting.
Authors: Armeet Singh Jatyani, Jiayun Wang, Aditi Chandrashekar, Zihui Wu, Miguel Liu-Schiaffini, Bahareh Tolooshams, Anima Anandkumar
Abstract: Compressed Sensing MRI reconstructs images of the body's internal anatomy from undersampled measurements, thereby reducing scan time. Recently, deep learning has shown great potential for reconstructing high-fidelity images from highly undersampled measurements. However, one needs to train multiple models for different undersampling patterns and desired output image resolutions, since most networks operate on a fixed discretization. Such approaches are highly impractical in clinical settings, where undersampling patterns and image resolutions are frequently changed to accommodate different real-time imaging and diagnostic requirements. We propose a unified MRI reconstruction model robust to various measurement undersampling patterns and image resolutions. Our approach uses neural operators, a discretization-agnostic architecture applied in both image and measurement spaces, to capture local and global features. Empirically, our model improves SSIM by 11% and PSNR by 4 dB over a state-of-the-art CNN (End-to-End VarNet), with 600$\times$ faster inference than diffusion methods. The resolution-agnostic design also enables zero-shot super-resolution and extended field-of-view reconstruction, offering a versatile and efficient solution for clinical MR imaging. Our unified model offers a versatile solution for MRI, adapting seamlessly to various measurement undersampling and imaging resolutions, making it highly effective for flexible and reliable clinical imaging. Our code is available at https://armeet.ca/nomri.
URLs: https://armeet.ca/nomri.
Authors: Pasan Dissanayake, Faisal Hamman, Barproda Halder, Ilia Sucholutsky, Qiuyi Zhang, Sanghamitra Dutta
Abstract: Knowledge distillation deploys complex machine learning models in resource-constrained environments by training a smaller student model to emulate internal representations of a complex teacher model. However, the teacher's representations can also encode nuisance or additional information not relevant to the downstream task. Distilling such irrelevant information can actually impede the performance of a capacity-limited student model. This observation motivates our primary question: What are the information-theoretic limits of knowledge distillation? To this end, we leverage Partial Information Decomposition to quantify and explain the transferred knowledge and knowledge left to distill for a downstream task. We theoretically demonstrate that the task-relevant transferred knowledge is succinctly captured by the measure of redundant information about the task between the teacher and student. We propose a novel multi-level optimization to incorporate redundant information as a regularizer, leading to our framework of Redundant Information Distillation (RID). RID leads to more resilient and effective distillation under nuisance teachers as it succinctly quantifies task-relevant knowledge rather than simply aligning student and teacher representations.
Authors: Amirhossein Abaskohi, Spandana Gella, Giuseppe Carenini, Issam H. Laradji
Abstract: Multimodal multihop question answering (MMQA) requires reasoning over images and text from multiple sources. Despite advances in visual question answering, this multihop setting remains underexplored due to a lack of quality datasets. Existing methods focus on single-hop, single-modality, or short texts, limiting real-world applications like interpreting educational documents with long, multimodal content. To fill this gap, we introduce FM2DS, the first framework for creating a high-quality dataset for MMQA. Our approach consists of a 5-stage pipeline that involves acquiring relevant multimodal documents from Wikipedia, synthetically generating high-level questions and answers, and validating them through rigorous criteria to ensure data quality. We evaluate our methodology by training models on our synthesized dataset and testing on two benchmarks: MultimodalQA and WebQA. Our results demonstrate that, with an equal sample size, models trained on our synthesized data outperform those trained on human-collected data by 1.9 in exact match (EM) score on average. Additionally, we introduce M2QA-Bench with 1k samples, the first benchmark for MMQA on long documents, generated using FM2DS and refined by human annotators. We believe our data synthesis method will serve as a strong foundation for training and evaluating MMQA models.
Authors: Max Getter
Abstract: Rapid information (energy) propagation in deep feature extractors is crucial to balance computational complexity versus expressiveness as a representation of the input. We prove an upper bound for the speed of energy propagation in a unified framework that covers different neural network models, both over Euclidean and non-Euclidean domains. Additional structural information about the signal domain can be used to explicitly determine or improve the rate of decay. To illustrate this, we show global exponential energy decay for a range of 1) feature extractors with discrete-domain input signals, and 2) convolutional neural networks (CNNs) via scattering over locally compact abelian (LCA) groups.
Authors: Evgeniia Vu, Andrei Boiarov, Dmitry Vetrov
Abstract: Generating co-speech gestures in real time requires both temporal coherence and efficient sampling. We introduce Accelerated Rolling Diffusion, a novel framework for streaming gesture generation that extends rolling diffusion models with structured progressive noise scheduling, enabling seamless long-sequence motion synthesis while preserving realism and diversity. We further propose Rolling Diffusion Ladder Acceleration (RDLA), a new approach that restructures the noise schedule into a stepwise ladder, allowing multiple frames to be denoised simultaneously. This significantly improves sampling efficiency while maintaining motion consistency, achieving up to a 2x speedup with high visual fidelity and temporal coherence. We evaluate our approach on ZEGGS and BEAT, strong benchmarks for real-world applicability. Our framework is universally applicable to any diffusion-based gesture generation model, transforming it into a streaming approach. Applied to three state-of-the-art methods, it consistently outperforms them, demonstrating its effectiveness as a generalizable and efficient solution for real-time, high-fidelity co-speech gesture synthesis.
Authors: Haoyang Wang, Ruishan Guo, Pengtao Ma, Ciyu Ruan, Xinyu Luo, Wenhua Ding, Tianyang Zhong, Jingao Xu, Yunhao Liu, Xinlei Chen
Abstract: With the increasing complexity of mobile device applications, these devices are evolving toward high agility. This shift imposes new demands on mobile sensing, particularly in terms of achieving high accuracy and low latency. Event-based vision has emerged as a disruptive paradigm, offering high temporal resolution, low latency, and energy efficiency, making it well-suited for high-accuracy and low-latency sensing tasks on high-agility platforms. However, the presence of substantial noisy events, the lack of inherent semantic information, and the large data volume pose significant challenges for event-based data processing on resource-constrained mobile devices. This paper surveys the literature over the period 2014-2024, provides a comprehensive overview of event-based mobile sensing systems, covering fundamental principles, event abstraction methods, algorithmic advancements, hardware and software acceleration strategies. We also discuss key applications of event cameras in mobile sensing, including visual odometry, object tracking, optical flow estimation, and 3D reconstruction, while highlighting the challenges associated with event data processing, sensor fusion, and real-time deployment. Furthermore, we outline future research directions, such as improving event camera hardware with advanced optics, leveraging neuromorphic computing for efficient processing, and integrating bio-inspired algorithms to enhance perception. To support ongoing research, we provide an open-source \textit{Online Sheet} with curated resources and recent developments. We hope this survey serves as a valuable reference, facilitating the adoption of event-based vision across diverse applications.
Authors: Takahiro Shirakawa, Tomoyuki Suzuki, Daichi Haraguchi
Abstract: General image-to-video generation methods often produce suboptimal animations that do not meet the requirements of animated graphics, as they lack active text motion and exhibit object distortion. Also, code-based animation generation methods typically require layer-structured vector data which are often not readily available for motion graphic generation. To address these challenges, we propose a novel framework named MG-Gen that reconstructs data in vector format from a single raster image to extend the capabilities of code-based methods to enable motion graphics generation from a raster image in the framework of general image-to-video generation. MG-Gen first decomposes the input image into layer-wise elements, reconstructs them as HTML format data and then generates executable JavaScript code for the reconstructed HTML data. We experimentally confirm that MG-Gen generates motion graphics while preserving text readability and input consistency. These successful results indicate that combining layer decomposition and animation code generation is an effective strategy for motion graphics generation.
Authors: Yan Ma, Steffi Chern, Xuyang Shen, Yiran Zhong, Pengfei Liu
Abstract: Reinforcement learning (RL) has recently shown strong potential in improving the reasoning capabilities of large language models and is now being actively extended to vision-language models (VLMs). However, existing RL applications in VLMs often rely on heavily engineered frameworks that hinder reproducibility and accessibility, while lacking standardized evaluation protocols, making it difficult to compare results or interpret training dynamics. This work introduces a transparent, from-scratch framework for RL in VLMs, offering a minimal yet functional four-step pipeline validated across multiple models and datasets. In addition, a standardized evaluation scheme is proposed to assess training dynamics and reflective behaviors. Extensive experiments on visual reasoning tasks uncover key empirical findings: response length is sensitive to random seeds, reflection correlates with output length, and RL consistently outperforms supervised fine-tuning (SFT) in generalization, even with high-quality data. These findings, together with the proposed framework, aim to establish a reproducible baseline and support broader engagement in RL-based VLM research.