CVPR 2025 论文列表

We introduce MMVU, a comprehensive expert-level, multi-discipline benchmark for evaluating foundation models in video understanding. MMVU includes 3,000 expert-annotated questions spanning 27 subjects across four core disciplines: Science, Healthcare, Humanities & Social Sciences, and Engineering. Compared to prior benchmarks, MMVU features three key advancements. First, it challenges models to apply domain-specific knowledge and perform expert-level reasoning to analyze specialized-domain videos, moving beyond the basic visual perception typically assessed in current video benchmarks. Second, each example is annotated by human experts from scratch. We implement strict data quality controls to ensure the high quality of the dataset. Finally, each example is enriched with expert-annotated reasoning rationals and relevant domain knowledge, facilitating in-depth analysis. We conduct an extensive evaluation of 36 frontier multimodal foundation models on MMVU. The latest System-2-capable models, o1 and Gemini 2.0 Flash Thinking, achieve the highest performance among the tested models. However, they still fall short of matching human expertise. Through in-depth error analyses and case studies, we offer actionable insights for future advancements in expert-level, knowledge-intensive video understanding for specialized domains.

Advanced driver assistance systems require a comprehensive understanding of the driver's mental/physical state and traffic context but existing works often neglect the potential benefits of joint learning between these tasks. This paper proposes MMTL-UniAD, a unified multi-modal multi-task learning framework that simultaneously recognizes driver behavior (e.g., looking around, talking), driver emotion (e.g., anxiety, happiness), vehicle behavior (e.g., parking, turning), and traffic context (e.g., traffic jam, traffic smooth). A key challenge is avoiding negative transfer between tasks, which can impair learning performance. To address this, we introduce two key components into the framework: one is the multi-axis region attention network to extract global context-sensitive features, and the other is the dual-branch multimodal embedding to learn multimodal embeddings from both task-shared and task-specific features. The former uses a multi-attention mechanism to extract task-relevant features, mitigating negative transfer caused by task-unrelated features. The latter employs a dual-branch structure to adaptively adjust task-shared and task-specific parameters, enhancing cross-task knowledge transfer while reducing task conflicts. We assess MMTL-UniAD on the AIDE dataset, using a series of ablation studies, and show that it outperforms state-of-the-art methods across all four tasks. The code is available on https://github.com/Wenzhuo-Liu/MMTL-UniAD.

Large-scale pre-trained Vision-Language Models (VLMs) have become essential for transfer learning across diverse tasks. However, adapting these models with limited few-shot data often leads to overfitting, diminishing their performance on new tasks. To tackle this issue, we propose a novel Multi-Modal Representation Learning (MMRL) framework that introduces a shared, learnable, and modality-agnostic representation space. MMRL projects the space tokens to text and image representation tokens, facilitating more effective multi-modal interactions. Unlike previous approaches that solely optimize class token features, MMRL integrates representation tokens at higher layers of the encoders--where dataset-specific features are more prominent--while preserving generalized knowledge in the lower layers. During training, both representation and class features are optimized, with trainable projection layer applied to the representation tokens, whereas the class token projection layer remains frozen to retain pre-trained knowledge. Furthermore, a regularization term is introduced to align the class features and text features with the zero-shot features from the frozen VLM, thereby safeguarding the model's generalization capacity. For inference, a decoupling strategy is employed, wherein both representation and class features are utilized for base classes, while only the class features, which retain more generalized knowledge, are used for new tasks. Extensive experiments across 15 datasets demonstrate that MMRL outperforms state-of-the-art methods, achieving a balanced trade-off between task-specific adaptation and generalization. Code is available at https://github.com/yunncheng/MMRL.

We propose to synthesize high-quality and synchronized audio, given video and optional text conditions, using a novel multimodal joint training framework (MMAudio). In contrast to single-modality training conditioned on (limited) video data only, MMAudio is jointly trained with larger-scale, readily available text-audio data to learn to generate semantically aligned high-quality audio samples. Additionally, we improve audio-visual synchrony with a conditional synchronization module that aligns video conditions with audio latents at the frame level. Trained with a flow matching objective, MMAudio achieves new video-to-audio state-of-the-art among public models in terms of audio quality, semantic alignment, and audio-visual synchronization, while having a low inference time (1.23s to generate an 8s clip) and just 157M parameters. MMAudio also achieves surprisingly competitive performance in text-to-audio generation, showing that joint training does not hinder single-modality performance. Code, models, and demo are available at: hkchengrex.github.io/MMAudio

Recent advancements in multi-modal large language models have propelled the development of joint probabilistic models capable of both image understanding and generation. However, we have identified that recent methods suffer from loss of image information during understanding task, due to either image discretization or diffusion denoising steps. To address this issue, we propose a novel Multi-Modal Auto-Regressive (MMAR) probabilistic modeling framework. Unlike discretization line of method, MMAR takes in continuous-valued image tokens to avoid information loss in an efficient way. Differing from diffusion-based approaches, we disentangle the diffusion process from auto-regressive backbone model by employing a light-weight diffusion head on top each auto-regressed image patch embedding. In this way, when the model transits from image generation to understanding through text generation, the backbone model's hidden representation of the image is not limited to the last denoising step. To successfully train our method, we also propose a theoretically proven technique that addresses the numerical stability issue and a training strategy that balances the generation and understanding task goals. Extensive evaluations on 18 image understanding benchmarks show that MMAR significantly outperforms most of the existing joint multi-modal models, surpassing the method that employs pre-trained CLIP vision encoder. Meanwhile, MMAR is able to generate high quality images. We also show that our method is scalable with larger data and model size.

Operating rooms (ORs) are complex, high-stakes environments requiring precise understanding of interactions among medical staff, tools, and equipment for enhancing surgical assistance, situational awareness, and patient safety. Current datasets fall short in scale, realism and do not capture the multimodal nature of OR scenes, limiting progress in OR modeling. To this end, we introduce MM-OR, a realistic and large-scale multimodal spatiotemporal OR dataset, and the first dataset to enable multimodal scene graph generation. MM-OR captures comprehensive OR scenes containing RGB-D data, detail views, audio, speech transcripts, robotic logs, and tracking data and is annotated with panoptic segmentations, semantic scene graphs, and downstream task labels. Further, we propose MM2SG, the first multimodal large vision-language model for scene graph generation, and through extensive experiments, demonstrate its ability to effectively leverage multimodal inputs. Together, MM-OR and MM2SG establish a new benchmark for holistic OR understanding, and open the path towards multimodal scene analysis in complex, high-stakes environments. We will publish all our code and dataset upon acceptance.

The evaluation of Long Video Understanding (LVU) performance poses an important but challenging research problem. Despite previous efforts, the existing video understanding benchmarks are severely constrained by several issues, especially the insufficient lengths of videos, a lack of diversity in video types and evaluation tasks, and the inappropriateness for evaluating LVU performances. To address the above problems, we propose a new benchmark called MLVU (Multi-task Long Video Understanding Benchmark) for the comprehensive and in-depth evaluation of LVU. MLVU presents the following critical values: 1) The substantial and flexible extension of video lengths, which enables the benchmark to evaluate LVU performance across a wide range of durations. 2) The inclusion of various video genres, such as movies, surveillance, egocentric videos, and cartoons, reflects the models' LVU performances in different scenarios. 3) The development of diversified evaluation tasks, which enables a comprehensive examination of MLLMs' key abilities in long-video understanding. The empirical study with 23 latest MLLMs reveals significant room for improvement in today's technique, as all existing methods struggle with most of the evaluation tasks and exhibit severe performance degradation when handling longer videos. Additionally, it suggests that factors such as context length, image-understanding ability, and the choice of LLM backbone can play critical roles in future advancements. We anticipate that MLVU will advance the research of LVU by providing a comprehensive and in-depth analysis of MLLMs. The code and dataset can be accessed from https://github.com/JUNJIE99/MLVU.

Image content safety has become a significant challenge with the rise of visual media on online platforms. Meanwhile, in the age of AI-generated content (AIGC), many image generation models are capable of producing harmful content, such as images containing sexual or violent material. Thus, it becomes crucial to identify such unsafe images based on established safety rules. Pre-trained Multimodal Large Language Models (MLLMs) offer potential in this regard, given their strong pattern recognition abilities. Existing approaches typically fine-tune MLLMs with human-labeled datasets, which however brings a series of drawbacks. First, relying on human annotators to label data following intricate and detailed guidelines is both expensive and labor-intensive. Furthermore, users of safety judgment systems may need to frequently update safety rules, making fine-tuning on human-based annotation more challenging. This raises the research question: Can we detect unsafe images by querying MLLMs in a zero-shot setting using a predefined safety constitution (a set of safety rules)? Our research showed that simply querying pre-trained MLLMs does not yield satisfactory results. This lack of effectiveness stems from factors such as the subjectivity of safety rules, the complexity of lengthy constitutions, and the inherent biases in the models. To address these challenges, we propose a MLLM-based method includes objectifying safety rules, assessing the relevance between rules and images, making quick judgments based on debiased token probabilities with logically complete yet simplified precondition chains for safety rules, and conducting more in-depth reasoning with cascaded chain-of-thought processes if necessary. Experiment results demonstrate that our method is highly effective for zero-shot image safety judgment tasks.

Multi-view object tracking (MVOT) offers promising solutions to challenges such as occlusion and target loss, which are common in traditional single-view tracking. However, progress has been limited by the lack of comprehensive multi-view datasets and effective cross-view integration methods. To overcome these limitations, we compiled a Multi-View object Tracking (MVTrack) dataset of 234K high-quality annotated frames featuring 27 distinct objects across various scenes. In conjunction with this dataset, we introduce a novel MVOT method, Multi-View Integration Tracker (MITracker), to efficiently integrate multi-view object features and provide stable tracking outcomes. MITracker can track any object in video frames of arbitrary length from arbitrary viewpoints. The key advancements of our method over traditional single-view approaches come from two aspects: (1) MITracker transforms 2D image features into a 3D feature volume and compresses it into a bird's eye view (BEV) plane, facilitating inter-view information fusion; (2) we propose an attention mechanism that leverages geometric information from fused 3D feature volume to refine the tracking results at each view. MITracker outperforms existing methods on the MVTrack and GMTD datasets, achieving state-of-the-art performance.

Implicit Neural Representations (INRs) are continuous function learners for conventional digital signal representations. With the aid of positional embeddings and/or exhaustively fine-tuned activation functions, INRs have surpassed many limitations of traditional discrete representations. However, existing works only find a continuous representation for the digital signal by solely using a single, fixed activation function throughout the INR, and it has not yet been explored to match the INR to the given signal. As current INRs are not matched to the signal being represented by the INR, we hypothesize that this approach could restrict the representation power and generalization capabilities of INRs, limiting their broader applicability. A way to match the INR to the signal being represented is through matching the activation of each layer in the sense of minimizing the mean squared loss. In this paper, we introduce MIRE, a method to find the highly matched activation function for each layer in INR through dictionary learning. To showcase the effectiveness of the proposed method, we utilize a dictionary that includes seven activation atoms: Raised Cosines (RC), Root Raised Cosines (RRC), Prolate Spheroidal Wave Function (PSWF), Sinc, Gabor Wavelet, Gaussian, and Sinusoidal. Experimental results demonstrate that MIRE not only significantly improves INR performance across various tasks, such as image representation, image inpainting, 3D shape representation, novel view synthesis, super-resolution, and reliable edge detection, but also eliminates the need for the previously required exhaustive search for activation parameters, which had to be conducted even before INR training could begin.

Character video synthesis aims to produce realistic videos of animatable characters within lifelike scenes. As a fundamental problem in the computer vision and graphics community, 3D works typically require multi-view captures for per-case training, which severely limits their applicability of modeling arbitrary characters in a short time. Recent 2D methods break this limitation via pre-trained diffusion models, but they struggle for flexible controls, pose generality and scene interaction. To this end, we propose MIMO, a novel framework which can not only synthesize realistic character videos with controllable attributes (i.e., character, motion and scene) provided by simple user inputs, but also simultaneously achieve advanced scalability to arbitrary characters, generality to novel 3D motions, and applicability to interactive real-world scenes in a unified framework. The core idea is to encode the 2D video to compact spatial codes, considering the inherent 3D nature of video occurrence. Concretely, we lift the 2D frame pixels into 3D using monocular depth estimators, and decompose the video clip into three spatial components (i.e., main human, underlying scene, and floating occlusion) in hierarchical layers based on the 3D depth. These components are further encoded to canonical identity code, structured motion code and full scene code, which are utilized as control signals of the synthesis process. The design of spatial decomposed modeling enables flexible user control, complex motion expression, as well as 3D-aware synthesis for scene interactions. Experimental results show that the proposed method outperforms prior works by a large margin in character animation synthesis and is effective in providing a high degree of controllability (i.e., arbitrary characters, novel 3D motions, interactive scenes), thus enabling brand-new editing tasks (e.g., video character replacement).

Currently, medical vision language models are widely used in medical vision question answering tasks. However, existing models are confronted with two issues: for input, the model only relies on text instructions and lacks direct understanding of visual clues in the image; for output, the model only gives text answers and lacks connection with key areas in the image. To address these issues, we propose a unified medical vision language model MIMO, with visual referring Multimodal Input and pixel grounding Multimodal Output. MIMO can not only combine visual clues and textual instructions to understand complex medical images and semantics, but can also ground medical terminologies in textual output within the image. To overcome the scarcity of relevant data in the medical field, we propose MIMOSeg, a comprehensive medical multimodal dataset including 895K samples. MIMOSeg is constructed from four different perspectives, covering basic instruction following and complex question answering with multimodal input and multimodal output. We conduct experiments on several downstream medical multimodal tasks. Extensive experimental results verify that MIMO can uniquely combine visual referring and pixel grounding capabilities, which are not available in previous models. Our project can be found in https://github.com/pkusixspace/MIMO.

This paper introduces MIDI, a novel paradigm for compositional 3D scene generation from a single image. Unlike existing methods that rely on reconstruction or retrieval techniques or recent approaches that employ multi-stage object-by-object generation, MIDI extends pre-trained image-to-3D object generation models to multi-instance diffusion models, enabling the simultaneous generation of multiple 3D instances with accurate spatial relationships and high generalizability. At its core, MIDI incorporates a novel multi-instance attention mechanism, that effectively captures inter-object interactions and spatial coherence directly within the generation process, without the need for complex multi-step processes. The method utilizes partial object images and global scene context as inputs, directly modeling object completion during 3D generation. During training, we effectively supervise the interactions between 3D instances using a limited amount of scene-level data, while incorporating single-object data for regularization, thereby maintaining the pre-trained generalization ability. MIDI demonstrates state-of-the-art performance in image-to-scene generation, validated through evaluations on synthetic data, real-world scene data, and stylized scene images generated by text-to-image diffusion models.

Point cloud processing (PCP) encompasses tasks like reconstruction, denoising, registration, and segmentation, each often requiring specialized models to address unique task characteristics. While in-context learning (ICL) has shown promise across tasks by using a single model with task-specific demonstration prompts, its application to PCP reveals significant limitations. We identify inter-task and intra-task sensitivity issues in current ICL methods for PCP, which we attribute to inflexible sampling strategies lacking context adaptation at the point and prompt levels. To address these challenges, we propose MICAS, an advanced ICL framework featuring a multi-grained adaptive sampling mechanism tailored for PCP. MICAS introduces two core components: task-adaptive point sampling, which leverages inter-task cues for point-level sampling, and query-specific prompt sampling, which selects optimal prompts per query to mitigate intra-task sensitivity. To our knowledge, this is the first approach to introduce adaptive sampling tailored to the unique requirements of point clouds within an ICL framework. Extensive experiments show that MICAS not only efficiently handles various PCP tasks but also significantly outperforms existing methods. Notably, it achieves a remarkable 4.1% improvement in the part segmentation task and delivers consistent gains across various PCP applications.

Based on analyzing the character of cascaded decoder architecture commonly adopted in existing DETR-like models, this paper proposes a new decoder architecture. The cascaded decoder architecture constrains object queries to update in the cascaded direction, only enabling object queries to learn relatively-limited information from image features. However, the challenges for object detection in natural scenes (e.g., extremely-small, heavily-occluded, and confusingly mixed with the background) require an object detection model to fully utilize image features, which motivates us to propose a new decoder architecture with the parallel Multi-time Inquiries (MI) mechanism. MI mechanism is very simple, enabling object queries to parallelly perform multi-time inquiries to learn more comprehensive information from image features. Our MI based model, MI-DETR, outperforms all existing DETR-like models on COCO benchmark under different backbones and training epochs, achieving +2.3 AP and +0.6 AP improvements compared to the most representative model DINO and SOTA model Relation-DETR under ResNet-50 backbone.

Recent motion-aware large language models have demonstrated promising potential in unifying motion comprehension and generation. However, existing approaches primarily focus on coarse-grained motion-text modeling, where text describes the overall semantics of an entire motion sequence in just a few words. This limits their ability to handle fine-grained motion-relevant tasks, such as understanding and controlling the movements of specific body parts. To overcome this limitation, we pioneer MG-MotionLLM, a unified motion-language model for multi-granular motion comprehension and generation. We further introduce a comprehensive multi-granularity training scheme by incorporating a set of novel auxiliary tasks, such as localizing temporal boundaries of motion segments via detailed text as well as motion detailed captioning, to facilitate mutual reinforcement for motion-text modeling across various levels of granularity. Extensive experiments show that our MG-MotionLLM achieves superior performance on classical text-to-motion and motion-to-text tasks, and exhibits potential in novel fine-grained motion comprehension and editing tasks. Project page: CVI-SZU/MG-MotionLLM

Deep learning approaches for marine fog detection and forecasting have outperformed traditional methods, demonstrating significant scientific and practical importance. However, the limited availability of open-source datasets remains a major challenge. Existing datasets, often focused on a single region or satellite, restrict the ability to evaluate model performance across diverse conditions and hinder the exploration of intrinsic marine fog characteristics. To address these limitations, we introduce MFogHub, the first multi-regional and multi-satellite dataset to integrate annotated marine fog observations from 15 coastal fog-prone regions and six geostationary satellites, comprising over 68,000 high-resolution samples. By encompassing diverse regions and satellite perspectives, MFogHub facilitates rigorous evaluation of both detection and forecasting methods under varying conditions. Extensive experiments with 16 baseline models demonstrate that MFogHub can reveal generalization fluctuations due to regional and satellite discrepancy, while also serving as a valuable resource for the development of targeted and scalable fog prediction techniques. Through MFogHub, we aim to advance both the practical monitoring and scientific understanding of marine fog dynamics on a global scale. The dataset and code are available in the supplementary materials.

Whole Slide Image (WSI) classification poses unique challenges due to the vast image size and numerous non-informative regions, which introduce noise and cause data imbalance during feature aggregation. To address these issues, we propose MExD, an Expert-Infused Diffusion Model that combines the strengths of a Mixture-of-Experts (MoE) mechanism with a diffusion model for enhanced classification. MExD balances patch feature distribution through a novel MoE-based aggregator that selectively emphasizes relevant information, effectively filtering noise, addressing data imbalance, and extracting essential features. These features are then integrated via a diffusion-based generative process to directly yield the class distribution for the WSI. Moving beyond conventional discriminative approaches, MExD represents the first generative strategy in WSI classification, capturing fine-grained details for robust and precise results. Our MExD is validated on three widely-used benchmarks--Camelyon16, TCGA-NSCLC, and BRACS--consistently achieving state-of-the-art performance in both binary and multi-class tasks.