CVPR 2025 Papers

Handling complex or nonlinear motion patterns has long posed challenges for video frame interpolation. Although recent advances in diffusion-based methods offer improvements over traditional optical flow-based approaches, they still struggle to generate sharp, temporally consistent frames in scenarios with large motion. To address this limitation, we introduce EDEN, an Enhanced Diffusion for high-quality large-motion vidEo frame iNterpolation. Our approach first utilizes a transformer-based tokenizer to produce refined latent representations of the intermediate frames for diffusion models. We then enhance the diffusion transformer with temporal attention across the process and incorporate a start-end frame difference embedding to guide the generation of dynamic motion. Extensive experiments demonstrate that EDEN achieves state-of-the-art results across popular benchmarks, including nearly a 10% LPIPS reduction on DAVIS and SNU-FILM, and an 8% improvement on DAIN-HD.

In Learned Video Compression (LVC), improving inter prediction, such as enhancing temporal context mining and mitigating accumulated errors, is crucial for boosting rate-distortion performance. Existing LVCs mainly focus on mining the temporal movements while neglecting non-local correlations among frames. Additionally, current contextual video compression models use a single reference frame, which is insufficient for handling complex movements. To address these issues, we propose leveraging non-local correlations across multiple frames to enhance temporal priors, significantly boosting rate-distortion performance. To mitigate error accumulation, we introduce a partial cascaded fine-tuning strategy that supports fine-tuning on full-length sequences with constrained computational resources. This method reduces the train-test mismatch in sequence lengths and significantly decreases accumulated errors. Based on the proposed techniques, we present a video compression scheme ECVC. Experiments demonstrate that our ECVC achieves state-of-the-art performance, reducing 10.5% and 11.5% more bit-rates than previous SOTA method DCVC-FM over VTM-13.2 low delay B (LDB) under the intra period (IP) of 32 and -1, respectively. Our code is available at https://github.com/JiangWeibeta/ECVC.

Blind Face Video Restoration (BFVR) focuses on reconstructing high-quality facial image sequences from degraded video inputs. The main challenge is address unknown degradations, while maintaining temporal consistency across frames. Current blind face restoration methods are primarily designed for images, and directly applying these approaches to BFVR will encounter a significant drop in restoration performance. In this work, we proposed Dynamic Content Prediction with Motion-aware Priors, referred to as DCP-MP. We develop a motion-aware semantic dictionary by encoding the semantic information of high-quality videos into discrete elements, and capturing the motion information in terms of element relationships, which are derived from the dynamic temporal changes within videos. For the purpose of utilizing dictionary to represent the degraded video, we train a temporal-aware element predictor, conditioned on degraded content, to learn the prediction of discrete elements in dictionary. The predicted elements will be refined, conditioned on motion information captured by the motion-aware semantic dictionary, to enhance temporal coherence. To alleviate deviation from the original structure information, we propose a conditional structure feature correction module that corrects the features flowing from the encoder to the generator. Through extensive experiments, we validate the effectiveness of our design components and demonstrate the superior performance of DCP-MP in synthesizing high-quality video.

The challenge in LLM-based video understanding lies in preserving visual and semantic information in long videos while maintaining a memory-affordable token count. However, redundancy and correspondence in videos have hindered the performance potential of existing methods. Through statistical learning on current datasets, we observe that redundancy occurs in both repeated and answer-irrelevant frames, and the corresponding frames vary with different questions. This suggests the possibility of adopting dynamic encoding to balance detailed video information preservation with token budget reduction. To this end, we propose a dynamic cooperative network, DynFocus, for memory-efficient video encoding in this paper. Specifically, i) a Dynamic Event Prototype Estimation (DPE) module to dynamically select meaningful frames for question answering; (ii) a Compact Cooperative Encoding (CCE) module that encodes meaningful frames with detailed visual appearance and the remaining frames with sketchy perception separately. We evaluate our method on five publicly available benchmarks, and experimental results consistently demonstrate that our method achieves competitive performance.

Humans can effortlessly locate desired objects in cluttered environments, relying on a cognitive mechanism known as visual search to efficiently filter out irrelevant information and focus on task related regions. Inspired by this process, we propose DyFo (Dynamic Focus), a training-free dynamic focusing visual search method that enhances fine-grained visual understanding in large multi-modal models (LMMs). Unlike existing approaches which require additional modules or data collection, DyFo leverages a bidirectional interaction between LMMs and visual experts, using a Monte Carlo Tree Search (MCTS) algorithm to simulate human-like focus adjustments. This enables LMMs to focus on key visual regions while filtering out irrelevant content, without introducing additional training caused by vocabulary expansion or the integration of specialized localization modules. Experimental results demonstrate that DyFo significantly improves fine-grained visual understanding and reduces hallucination issues in LMMs, achieving superior performance across both fixed and dynamic resolution models.

Video large language models (VLLMs) have significantly advanced recently in processing complex video content. Yet, their inference efficiency remains constrained because of the high computational cost stemming from the thousands of visual tokens generated from the video inputs. We empirically observe that, unlike single image inputs, VLLMs typically attend visual tokens from different frames at different decoding iterations. This makes a one-shot pruning strategy prone to removing important tokens by mistake. Motivated by this, we present DyCoke, a training-free token compression method to optimize token representation and accelerate VLLMs. DyCoke incorporates a plug-and-play temporal compression module to minimize temporal redundancy by merging redundant tokens across frames and applying dynamic KV cache reduction to prune spatially redundant tokens selectively. It ensures high-quality inference by dynamically retaining the critical tokens at each decoding step. Extensive experimental results demonstrate that DyCoke can outperform the prior SoTA counterparts, achieving 1.5x inference speedup, and 1.4x memory reduction against the baseline VLLM, while still improving the performance, with no training.

Most of the existing methods for video understanding primarily focus on videos only lasting tens of seconds, with limited exploration of techniques for handling long videos. The increased number of frames in long videos poses two main challenges: difficulty in locating key information and performing long-range reasoning. Thus, we propose DrVideo, a document-retrieval-based system designed for long video understanding. Our key idea is to convert the long-video understanding problem into a long-document understanding task so as to effectively leverage the power of large language models. Specifically, DrVideo first transforms a long video into a coarse text-based long document to initially retrieve key frames and then updates the documents with the augmented key frame information. It then employs an agent-based iterative loop to continuously search for missing information and augment the document until sufficient question-related information is gathered for making the final predictions in a chain-of-thought manner. Extensive experiments on long video benchmarks confirm the effectiveness of our method. DrVideo significantly outperforms existing LLM-based state-of-the-art methods on EgoSchema benchmark (3 minutes), MovieChat-1K benchmark (10 minutes), and the long split of Video-MME benchmark (average of 44 minutes). Code is available at https://github.com/Upper9527/DrVideo.

Text-rich document understanding (TDU) requires comprehensive analysis of documents containing substantial textual content and complex layouts. While Multimodal Large Language Models (MLLMs) have achieved fast progress in this domain, existing approaches either demand significant computational resources or struggle with effective multi-modal integration. In this paper, we introduce DocLayLLM, an efficient multi-modal extension of LLMs specifically designed for TDU. By lightly integrating visual patch tokens and 2D positional tokens into LLMs' input and encoding the document content using the LLMs themselves, we fully take advantage of the document comprehension capability of LLMs and enhance their perception of OCR information. We have also deeply considered the role of chain-of-thought (CoT) and innovatively proposed the techniques of CoT Pre-training and CoT Annealing. Our DocLayLLM can achieve remarkable performances with lightweight training settings, showcasing its efficiency and effectiveness. Experimental results demonstrate that our DocLayLLM outperforms existing OCR-dependent methods and OCR-free competitors. Code and model are available at https://github.com/whlscut/DocLayLLM.

Active Real-time interaction with video LLMs introduces a new paradigm for human-computer interaction, where the model not only understands user intent but also responds while continuously processing streaming video on the fly. Unlike offline video LLMs, which analyze the entire video before answering questions, active real-time interaction requires three capabilities: 1) Perception: real-time video monitoring and interaction capturing. 2) Decision: raising proactive interaction in proper situations, 3) Reaction: continuous interaction with users. However, inherent conflicts exist among the desired capabilities. The Decision and Reaction require a contrary Perception scale and grain, and the autoregressive decoding blocks the real-time Perception and Decision during the Reaction. To unify the conflicted capabilities within a harmonious system, we present Dispider, a solution built on a Disentangled Perception, Decision, and Reaction framework. Dispider features a lightweight Proactive Streaming Video Processing module that tracks the video stream and identifies optimal moments for interaction. Once the interaction is triggered, an asynchronous Precise Interaction module provides detailed responses, while the processing module continues to monitor the video in the meantime. Our disentangled and asynchronous design ensures timely, contextually accurate, and computationally efficient responses, making Dispider ideal for active real-time interaction for long-duration video streams. Experiments prove that Dispider outperforms existing methods not only in its superior understanding of video content in conventional video QA settings, but also in proactive response capability and temporal awareness under the streaming setting.

The parameter-efficient adaptation of the image-text pretraining model CLIP for video-text retrieval is a prominent area of research. While CLIP is focused on image-level vision-language matching, video-text retrieval demands comprehensive understanding at the video level. Three key discrepancies emerge in the transfer from image-level to video-level: vision, language, and alignment. However, existing methods mainly focus on vision while neglecting language and alignment. In this paper, we propose Discrepancy Reduction in Vision, Language, and Alignment (DiscoVLA), which simultaneously mitigates all three discrepancies. Specifically, we introduce Image-Video Features Fusion to integrate image-level and video-level features, effectively tackling both vision and language discrepancies. Additionally, we generate pseudo image captions to learn fine-grained image-level alignment. To mitigate alignment discrepancies, we propose Image-to-Video Alignment Distillation, which leverages image-level alignment knowledge to enhance video-level alignment. Extensive experiments demonstrate the superiority of our DiscoVLA. In particular, on MSRVTT with CLIP (ViT-B/16), DiscoVLA outperforms previous methods by 2.2% R@1 and 7.5% R@sum. The code is available at https://github.com/LunarShen/DsicoVLA.

Understanding and modeling lighting effects are fundamental tasks in computer vision and graphics. Classic physically-based rendering (PBR) accurately simulates the light transport, but relies on precise scene representations--explicit 3D geometry, high-quality material properties, and lighting conditions--that are often impractical to obtain in real-world scenarios. Therefore, we introduce Diffusion Renderer, a neural approach that addresses the dual problem of inverse and forward rendering within a holistic framework. Leveraging powerful video diffusion model priors, the inverse rendering model accurately estimates G-buffers from real-world videos, providing an interface for image editing tasks, and training data for the rendering model. Conversely, our rendering model generates photorealistic images from G-buffers without explicit light transport simulation. Specifically, we first train a video diffusion model for inverse rendering on synthetic data, which generalizes well to real-world videos and allows us to auto-label diverse real-world videos. We then co-train our rendering model using both synthetic and auto-labeled real-world data. Experiments demonstrate that Diffusion Renderer effectively approximates inverse and forwards rendering, consistently outperforming the state-of-the-art. Our model enables practical applications from a single video input--including relighting, material editing, and realistic object insertion.

Top-leading solutions for Video Scene Graph Generation (VSGG) typically adopt an offline pipeline.Though demonstrating promising performance, they remain unable to handle real-time video streams and consume large GPU memory. Moreover, these approaches fall short in temporal reasoning, merely aggregating frame-level predictions over a temporal context. In response, we introduce DiffVsgg, an online VSGG solution that frames this task as an iterative scene graph update problem. Drawing inspiration from Latent Diffusion Models (LDMs) which generate images via denoising a latent feature embedding, we unify the decoding of object classification, bounding box regression, and graph generation three tasks using one shared feature embedding. Then, given an embedding containing unified features of object pairs, we conduct a step-wise Denoising on it within LDMs, so as to deliver a clean embedding which clearly indicates the relationships between objects.This embedding then serves as the input to task-specific heads for object classification, scene graph generation, etc. DiffVsgg further facilitates continuous temporal reasoning, where predictions for subsequent frames leverage results of past frames as the conditional inputs of LDMs, to guide the reverse diffusion process for current frames.Extensive experiments on three setups of Action Genome demonstrate the superiority of DiffVsgg.

In recent years, large transformer-based video encoder models have greatly advanced state-of-the-art performance on video classification tasks. However, these large models typically process videos by averaging embedding outputs from multiple clips over time to produce fixed-length representations. This approach fails to account for a variety of time-related features, such as variable video durations, chronological order of events, and temporal variance in feature significance. While methods for temporal modeling do exist, they often require significant architectural changes and expensive retraining, making them impractical for off-the-shelf, fine-tuned large encoders. To overcome these limitations, we propose DejaVid, an encoder-agnostic method that enhances model performance without the need for retraining or altering the architecture. Our framework converts a video into a variable-length temporal sequence of embeddings (TSE). A TSE naturally preserves temporal order and accommodates variable video durations. We then learn per-timestep, per-feature weights over the encoded TSE frames, allowing us to account for variations in feature importance over time. We introduce a new neural network architecture inspired by traditional time series alignment algorithms for this learning task. Our evaluation demonstrates that DejaVid substantially improves the performance of a state-of-the-art large encoder, achieving leading Top-1 accuracy of 77.2% on Something-Something V2, 89.1% on Kinetics-400, and 88.6% on HMDB51, while adding fewer than 1.8% additional learnable parameters and requiring less than 3 hours of training time. Our code is available at https://github.com/darrylho/DejaVid.

Motion expression video segmentation aims to segment objects based on input motion descriptions. Compared with traditional referring video object segmentation, it focuses on motion and multi-object expressions and is more challenging. Previous works achieved it by simply injecting text information into the video instance segmentation (VIS) model. However, this requires retraining the entire model and optimization is difficult. In this work, we propose DMVS, a simple framework constructed on the existing query-based VIS model, emphasizing decoupling the task into video instance segmentation and motion expression understanding. Firstly, we use a frozen video instance segmenter to extract object-specific contexts and convert them into frame-level and video-level queries. Secondly, we interact two levels of queries with static and motion cues, respectively, to further encode visually enhanced motion expressions. Furthermore, we propose a novel query initialization strategy that uses video queries guided by classification priors to initialize motion queries, greatly reducing the difficulty of optimization. Without bells and whistles, DMVS achieves state-of-the-art performance on the MeViS dataset at a lower training cost. Extensive experiments verify the effectiveness and efficiency of our framework.

We introduce \text D ^3\text -Human , a method for reconstructing Dynamic Disentangled Digital Human geometry from monocular videos. Past monocular video human reconstruction primarily focuses on reconstructing undecoupled clothed human bodies or only reconstructing clothing, making it difficult to apply directly in applications such as animation production. The challenge in reconstructing decoupled clothing and body lies in the occlusion caused by clothing over the body. To this end, the details of the visible area and the plausibility of the invisible area must be ensured during the reconstruction process. Our proposed method combines explicit and implicit representations to model the decoupled clothed human body, leveraging the robustness of explicit representations and the flexibility of implicit representations. Specifically, we reconstruct the visible region as SDF and propose a novel human manifold signed distance field (hmSDF) to segment the visible clothing and visible body, and then merge the visible and invisible body. Extensive experimental results demonstrate that, compared with existing reconstruction schemes, \text D ^3\text -Human can achieve high-quality decoupled reconstruction of the human body wearing different clothing, and can be directly applied to clothing transfer and animation production.

Environment understanding in egocentric videos is an important step for applications like robotics, augmented reality and assistive technologies. These videos are characterized by dynamic interactions and a strong dependence on the wearer's engagement with the environment. Traditional approaches often focus on isolated clips or fail to integrate rich semantic and geometric information, limiting scene comprehension. We introduce Dynamic Image-Video Feature Fields (DIV-FF), a framework that decomposes the egocentric scene into persistent, dynamic, and actor-based components while integrating both image and video-language features. Our model enables detailed segmentation, captures affordances, understands the surroundings and maintains consistent understanding over time. DIV-FF outperforms state-of-the-art methods, particularly in dynamically evolving scenarios, demonstrating its potential to advance long-term, spatio-temporal scene understanding.

Video question grounding (VideoQG) requires models to answer the questions and simultaneously infer the relevant video segments to support the answers. However, existing VideoQG methods usually suffer from spurious cross-modal correlations, leading to a failure to identify the dominant visual scenes that align with the intended question. Moreover, vision-language models exhibit unfaithful generalization performance and lack robustness on challenging downstream tasks such as VideoQG. In this work, we propose a novel VideoQG framework named Cross-modal Causal Relation Alignment (CRA), to eliminate spurious correlations and improve the causal consistency between question-answering and video temporal grounding. Our CRA involves three essential components: i) Gaussian Smoothing Grounding (GSG) module for estimating the time interval via cross-modal attention, which is de-noised by an adaptive Gaussian filter, ii) Cross-Modal Alignment (CMA) enhances the performance of weakly supervised VideoQG by leveraging bidirectional contrastive learning between estimated video segments and QA features, iii) Explicit Causal Intervention (ECI) module for multimodal deconfounding, which involves front-door intervention for vision and back-door intervention for language. Extensive experiments on two VideoQG datasets demonstrate the superiority of our CRA in discovering visually grounded content and achieving robust question reasoning. Codes are available at https://github.com/WissingChen/CRA-GQA.

The lack of a large-scale 3D-text corpus has led recent works to distill open-vocabulary knowledge from vision-language models (VLMs). However, these methods typically rely on a single VLM to align the feature spaces of 3D models within a common language space, which limits the potential of 3D models to leverage the diverse spatial and semantic capabilities encapsulated in various foundation models. In this paper, we propose Cross-modal and Uncertainty-aware Agglomeration for Open-vocabulary 3D Scene Understanding dubbed CUA-O3D, the first model to integrate multiple foundation models--such as CLIP, DINOv2, and Stable Diffusion--into 3D scene understanding. We further introduce a deterministic uncertainty estimation to adaptively distill and harmonize the heterogeneous 2D feature embeddings from these models. Our method addresses two key challenges: (1) incorporating semantic priors from VLMs alongside the geometric knowledge of spatially-aware vision foundation models, and (2) using a novel deterministic uncertainty estimation to capture model-specific uncertainties across diverse semantic and geometric sensitivities, helping to reconcile heterogeneous representations during training. Extensive experiments on ScanNetV2 and Matterport3D demonstrate that our method not only advances open-vocabulary segmentation but also achieves robust cross-domain alignment and competitive spatial perception capabilities.

Space-time video resampling aims to conduct both spatial-temporal downsampling and upsampling processes to achieve high-quality video reconstruction.Although there has been much progress, some major challenges still exist, such as how to preserve motion information during temporal resampling while avoiding blurring artifacts, and how to achieve flexible temporal and spatial resampling factors. In this paper, we introduce an Invertible Motion Steganography Module (IMSM), designed to preserve motion information in high-frame-rate videos. This module embeds motion information from high-frame-rate videos into downsampled frames with lower frame rates in a visually imperceptible manner. Its reversible nature allows the motion information to be recovered, facilitating the reconstruction of high-frame-rate videos. Furthermore, we propose a 3D implicit feature modulation technique that enables continuous spatiotemporal resampling. With tailored training strategies, our method supports flexible frame rate conversions, including non-integer changes like 30 FPS to 24 FPS and vice versa. Extensive experiments show that our method significantly outperforms existing solutions across multiple datasets in various video resampling tasks with high flexibility. Codes will be made available.