CVPR 2026 论文列表

Causality--referring to temporal, uni-directional cause-effect relationships between components--underlies many complex generative processes, including videos, language, and robot trajectories.Current causal diffusion models entangle temporal reasoning with iterative denoising, applying causal attention across all layers, at every denoising step, and over the entire context.In this paper, we show that the causal computation in these models is separable from the multi-step denoising process.Through systematic probing of autoregressive video diffusers, we uncover two key regularities:(1) early blocks produce highly similar features across denoising steps, indicating redundant computation along the diffusion trajectory; and(2) deeper blocks exhibit sparse cross-frame attention and primarily perform intra-frame rendering.Motivated by these findings, we introduce Separable Causal Diffusion (SCD), a new architecture that explicitly decouples once-per-frame temporal reasoning, via a causal transformer encoder, from multi-step frame-wise rendering, via a lightweight diffusion decoder.Extensive experiments on both pretraining and post-training tasks across synthetic and real benchmarks show that CSD significantly improves throughput and latency while matching or surpassing the generation quality of strong causal diffusion baselines.

Frame selectoin is crucial due to high frame redundancy and limited context windows when applying Large Vision-Language Models (LVLMs) to long videos. Current methods typically select frames with high relevance to a given query, resulting a disjointed set of frames that disregard the narrative structure of video. In this paper, we introduce Wavelet-based Frame Selection by Detecting Semantic Boundary (WFS-SB), a training-free framework that presents a new perspective: effective video understanding hinges not only on high relevance but, more importantly, on capturing semantic shifts--pivotal moments of narrative change that are essential to comprehending the holistic storyline of video. However, a direct detection of abrupt changes in the query-frame similarity signal is often unreliable due to high-frequency noise arising from model uncertainty and transient visual variations.To address this, we leverage the wavelet transform, which provides an ideal solution through its multi-resolution analysis in both time and frequency domains. By applying this transform, we decompose the noisy signal into multiple scales and extract a clean semantic change signal from the coarsest scale. We identify the local extrema of this signal as semantic boundaries, which segment the video into coherent clips. Building on this, WFS-SB comprises a two-stage strategy: first, adaptively allocating a frame budget to each clip based on a composite importance score; and second, within each clip, employing the Maximal Marginal Relevance approach to select a diverse yet relevant set of frames. Extensive experiments show that WFS-SB significantly boosts LVLM performance, e.g., improving accuracy by 5.5% on VideoMME, 9.5% on MLVU, and 6.2% on LongVideoBench, consistently outperforming state-of-the-art methods. Our code is available at https://github.com/MAC-AutoML/WFS-SB.

We propose a method to reconstruct high-fidelity human avatars from multi-view video that can run on mobile devices. Many works can model high-quality Gaussian-based full-body avatars from multi-view video. However, these methods require heavy computation to obtain pose-dependent appearance, making deployment on mobile devices very difficult. Recent methods distill from pretrained models and model pose-dependent nonlinear Gaussian attributes by linearly combining global pose features with blendshapes. Although they can run on mobile devices, they suffer some loss of detail. We observe that nearby Gaussians are often highly correlated within a local region of the body, and can be linearly modeled with less error. Therefore, we use local linear blendshapes in small body parts to capture global nonlinear changes of Gaussian attributes. To further reduce computation and model size, we propose to remove blendshapes for Gaussians whose attributes change little, yielding a minimal blendshape representation. Our method is an end-to-end training method without a pretrained model. To make it run on multiple devices, we implement our method using WebGPU. Experiments show that our method can render high-quality human avatars with better details, and can reach 120 FPS at 2K resolution on mobile devices.

Learning temporal transformations, that is, how visual objects evolve across frames, is a fundamental challenge in video representation learning. Frame-to-frame dynamics involve complex, non-linear, and non-local changes that go far beyond conventional spatial augmentations. We propose TimeBridge, a self-supervised method that combines the joint embedding for video representation with learning temporal transformations by reconstructing in-between frames from only the start and end frames. This formulation encourages the model to infer the temporal evolution bridging the two endpoints, rather than merely encoding static frame representations. Unlike joint-embedding methods that lack explicit transformation modelling or future-prediction objectives that rely on unconstrained extrapolation, TimeBridge learns concrete frame-to-frame dynamics by promoting temporal consistency. We realise this through cross-concatenated class tokens and lightweight decoders, which recombine features from the start and end frames to reconstruct intermediates. TimeBridge achieves new state-of-the-art performance on multiple dense video prediction benchmarks, including 73.5 J&F on DAVIS 2017 video object segmentation, 47.5 mIoU on VIP semantic part propagation.

Streaming Video Large Language Models (VideoLLMs) have demonstrated impressive performance across various video understanding tasks, but they face significant challenges in real-time deployment due to the high computational cost of processing dense visual tokens from continuous video streams. In streaming video scenarios, the primary bottleneck lies in the Vision Transformer (ViT) encoding stage, where redundant processing of temporally similar frames leads to inefficiency. Additionally, inflated token sequences during LLM pre-filling further exacerbate latency and memory overhead. To address these challenges, we propose Streaming Token Compression (STC), a plug-and-play hierarchical framework that seamlessly integrates into existing streaming VideoLLMs, optimizing both ViT encoding and LLM pre-filling stages to accelerate processing. STC introduces two token-level accelerators: STC-Cacher, which reduces ViT encoding overhead by caching and reusing features from temporally similar frames, and STC-Pruner, which compresses the visual token sequence before it enters the LLM, preserving only the most salient tokens based on both spatial and temporal relevance. Extensive experiments on four baseline streaming VideoLLMs across five benchmarks demonstrate that STC outperforms other compression methods. Notably, STC retains up to 99% of accuracy on the ReKV framework while reducing ViT encoding latency and LLM pre-filling latency by 24.5% and 45.3%.

The new era has witnessed a remarkable capability to extend Vision-Language Models (VLMs) for tackling tasks of video understanding. While current VLMs excel at event- or story-level understanding, their ability to capture fine-grained motion details remains limited, primarily due to their focus on high-level static semantic structures and macro-event logic. In contrast, Video Diffusion Models (VDMs) are adept at modeling dynamic motion patterns, benefiting from large-scale video data and the intrinsic requirement of temporal generation. In this paper, we introduce MotionEnhancer, a novel approach that leverages motion priors distilled from a powerful video diffusion model as auxiliary supervision to enhance the motion understanding capability of a VLM via attention alignment. MotionEnhancer comprises two simple parameter-free modules, Motion-sensitive Head Selection (MHS) and Motion-salient Text Token Identification (MTTI), to directly extract and optimize motion-related attentions from the VDM in a computation-only manner. MotionEnhancer provides a scalable solution for motion understanding without additional training parameters, modifications to existing architectures, or tool calling. Extensive experiments demonstrate that MotionEnhancer can achieve consistent improvements over state-of-the-art VLMs on two motion-level video understanding benchmarks, especially on motion-related metrics.

Text-video retrieval tasks have seen significant improvements due to the recent development of large-scale vision-language pre-trained models. Traditional methods primarily focus on video representations or cross-modal alignment, while recent works shift toward enriching text expressiveness to better match the rich semantics in videos. However, these methods use only interactions between text and frames/video, and ignore rich interactions among the internal frames within a video, so the final expanded text cannot capture frame contextual information, leading to disparities between text and video. In response, we introduce Energy-Aware Fine-Grained Relationship Learning Network (EagleNet) to generate accurate and context-aware enriched text embeddings. Specifically, the proposed Fine-Grained Relationship Learning mechanism (FRL) first constructs a text-frame graph by the generated text candidates and frames, then learns relationships among texts and frames, which are finally used to aggregate text candidates into an enriched text embedding that incorporates frame contextual information. To further improve fine-grained relationship learning in FRL, we design Energy-Aware Matching (EAM) to model the energy of text-frame interactions and thus accurately capture the distribution of real text-video pairs. Moreover, for more effective cross-modal alignment and stable training, we replace the conventional softmax-based contrastive loss with the sigmoid loss. Extensive experiments have demonstrated the superiority of EagleNet across MSRVTT, DiDeMo, MSVD, and VATEX.

Video Temporal Grounding (VTG) aims to localize the video segment that corresponds to a natural language query, which requires a comprehensive understanding of complex temporal dynamics. Existing Vision-LMMs typically perceive temporal dynamics via positional encoding, text-based timestamps, or visual frame numbering. However, these approaches exhibit notable limitations: assigning each frame a text-based timestamp token introduces additional computational overhead and leads to sparsity in visual attention, positional encoding struggles to capture absolute temporal information, and visual frame numbering often compromises spatial detail. To address these issues, we propose Temporal to Spatial Gridification (T2SGrid), a novel framework that reformulates video temporal understanding as a spatial understanding task. The core idea of T2SGrid is to process video content in clips rather than individual frames. we employ a overlapping sliding windows mechanism to segment the video into temporal clips. Within each window, frames are arranged chronologically in a row-major order into a composite grid image, effectively transforming temporal sequences into structured 2D layouts. The gridification not only encodes temporal information but also enhances local attention within each grid. Furthermore, T2SGrid enables the use of composite text timestamps to establish global temporal awareness. Experiments on standard VTG benchmarks demonstrate that T2SGrid achieves superior performance.

Humans develop visual intelligence through perceiving and interacting with their environment--a self-supervised learning process grounded in egocentric experience. Inspired by this, we ask how can artificial systems learn stable object representations from continuous, uncurated first-person videos without relying on manual annotations. This setting poses challenges of separating, recognizing, and persistently tracking objects amid clutter, occlusion, and ego-motion. We propose EgoViT, a unified vision Transformer framework designed to learn stable object representations from unlabeled egocentric video. EgoViT bootstraps this learning process by jointly discovering and stabilizing "proto-objects" through three synergistic mechanisms: (1) Proto-object Learning, which uses intra-frame distillation to form discriminative representations; (2) Depth Regularization, which grounds these representations in geometric structure; and (3) Teacher-Filtered Temporal Consistency, which enforces identity over time. This creates a virtuous cycle where initial object hypotheses are progressively refined into stable, persistent representations. The framework is trained end-to-end on unlabeled first-person videos and exhibits robustness to geometric priors of varied origin and quality. On standard benchmarks, EgoViT achieves +8.0% CorLoc improvement in unsupervised object discovery and +4.8% mIoU improvement in semantic segmentation, demonstrating its potential to lay a foundation for robust visual abstraction in embodied intelligence.

Large vision foundation models, such as SAM2, have achieved remarkable performance in video object segmentation and tracking (VOST). However, their effectiveness is hindered by significant computational overhead. While model pruning is a widely used strategy to address this issue, traditional static and input-agnostic pruning approaches fall short in managing the diverse and complex nature of video data effectively. A promising alternative is dynamic networks, yet they often struggle to translate theoretical computational reductions into actual acceleration. Furthermore, both static and dynamic approaches typically focus on visual features of individual frames while neglecting the temporal correlations between them, limiting their performance in handling complex video streams.To address these challenges, we propose Recurrent Dynamic Submodel (RDS), a dynamic architecture that adaptively selects submodel blocks for each frame. Specifically, it has a lightweight Prediction-aware Router (PAR), which leverages both the segmentation mask from the previous frame and the visual features of the current frame to make routing decisions, enabling the submodel to explicitly capture the temporal nature of video data. Additionally, to reduce the cost of adapting the dynamic submodel, we introduce an Importance-aware LoRA (I-LoRA), tuning parameters only in the most critical blocks. Extensive experiments show that our approach achieves an optimal trade-off between performance and speed, requiring minimal trainable parameters and training data.

Clinical video diagnosis, in which physicians assess dynamic tissue responses across procedural stages, is critical for detecting diseases such as cervical and colorectal cancers. Recent spatiotemporal models map visual progressions directly to diagnostic outputs, yet overlook two hallmarks of expert reasoning: clinically grounded diagnostic principles and hypothesis-driven counterfactual thinking. This gap leads existing methods to conflate causal pathological cues with non-pathological variations, thereby limiting their reliability in data-scarce settings. Here we show that explicitly modeling counterfactual tissue evolution, guided by clinical rules that encode expert diagnostic principles, substantially improves diagnostic robustness-emulating the hypothesis-driven reasoning clinicians employ in practice. We introduce MEDVCR, comprising a Counterfactual Generator (CG) that synthesizes hypothetical tissue transitions via diffusion modeling, a Counterfactual Representation Learning (CRL) module that enforces temporal consistency, pathological separability, and counterfactual alignment, and a Dual Diagnostic Prediction (DDP) strategy that combines video-level context with frame-level counterfactual contrast. On two representative tasks, MEDVCR achieves 93.0% Recall@1 on colposcopy (+10.2%) and 94.8% Average Precision (AP) on colonoscopy (+2.6%), outperforming all state-of-the-art (SOTA) baselines. We anticipate that this counterfactual reasoning paradigm will open new avenues for vision-based clinical diagnostic tasks toward more transparent and interpretable decision support.

Long video understanding is essential for human-like intelligence, enabling coherent perception and reasoning over extended temporal contexts. While the emerging thinking-with-frames paradigm--which alternates between global temporal reasoning and local frame examination--has advanced the reasoning capabilities of video multi-modal large language models (MLLMs), it suffers from a significant efficiency bottleneck due to the progressively growing and redundant multi-modal context. To address this, we propose SpecTemp, a reinforcement learning-based Speculative Temporal reasoning framework that decouples temporal perception from reasoning via a cooperative dual-model design. In SpecTemp, a lightweight draft MLLM rapidly explores and proposes salient frames from densely sampled temporal regions, while a powerful target MLLM focuses on temporal reasoning and verifies the draft's proposals, iteratively refining its attention until convergence. This design mirrors the collaborative pathways of the human brain, balancing efficiency with accuracy. To support training, we construct the SpecTemp-80K dataset, featuring synchronized dual-level annotations for coarse evidence spans and fine-grained frame-level evidence. Experiments across multiple video understanding benchmarks demonstrate that SpecTemp not only maintains competitive accuracy but also significantly accelerates inference compared with existing thinking-with-frames methods.

Unified large multimodal models (LMMs) have achieved remarkable progress in general-purpose multimodal understanding and generation. However, they still operate under a "one-size-fits-all" paradigm and struggle to model user-specific concepts (e.g., generate a photo of \texttt \<maeve> ) in a consistent and controllable manner. Existing personalization methods typically rely on external retrieval, which is inefficient and poorly integrated into unified multimodal pipelines. Recent personalized unified models introduce learnable soft prompts to encode concept information, yet they either couple understanding and generation or depend on complex multi-stage training, leading to cross-task interference and ultimately to fuzzy or misaligned personalized knowledge.We present OmniPersona, an end-to-end personalization framework for unified LMMs that, for the first time, integrates personalized understanding, generation, and image editing within a single architecture. OmniPersona introduces structurally decoupled concept tokens, allocating dedicated subspaces for different tasks to minimize interference, and incorporates an explicit knowledge replay mechanism that propagates personalized attribute knowledge across tasks, enabling consistent personalized behavior.To systematically evaluate unified personalization, we propose \texttt OmniPBench , extending the public UnifyBench concept set with personalized editing tasks and cross-task evaluation protocols integrating understanding, generation, and editing. Experimental results demonstrate that OmniPersona delivers competitive and robust performance across diverse personalization tasks. We hope OmniPersona will serve as a strong baseline and spur further research on controllable, unified personalization.

Fully supervised Video Semantic Segmentation (VSS) relies heavily on densely annotated video data, limiting practical applicability. Alternatively, applying pre-trained Image Semantic Segmentation (ISS) models frame-by-frame avoids annotation costs but ignores crucial temporal coherence. Recent foundation models such as SAM2 enable high-quality mask propagation yet remain impractical for direct VSS due to limited semantic understanding and computational overhead. In this paper, we propose DiTTA (Distillation-assisted Test-Time Adaptation), a novel framework that converts an ISS model into a temporally-aware VSS model through efficient test-time adaptation (TTA), without annotated videos. DiTTA distills SAM2's temporal segmentation knowledge into the ISS model during a brief, single-pass initialization phase, complemented by a lightweight temporal fusion module to aggregate cross-frame context. Crucially, DiTTA achieves robust generalization even when adapting with highly limited partial video snippets (e.g., initial 10%), significantly outperforming zero-shot refinement approaches that repeatedly invoke SAM2 during inference. Extensive experiments on VSPW and Cityscapes demonstrate DiTTA's effectiveness, achieving competitive or superior performance relative to fully-supervised VSS methods, thus providing a practical and annotation-free solution for real-world VSS tasks.

Crime anticipation enables proactive public safety interventions, yet existing video security systems remain largely reactive, unable to detect precursors of crime. While current visual language models (VLM)-based video understanding methods show promise in high-level reasoning, they are not designed to explicitly model the spatio-temporal causal relationships essential for anticipating crimes. We address this limitation by two causal-driven components. First, we develop the Spatio-Temporal Causal Reasoning Crime (STCRC) dataset, a hierarchical dataset comprising 73K samples across five progressive causal reasoning tasks, facilitating criminal precursors learning. Second, we propose the Spatio-Temporal Causal Hypergraph (STCH), a streaming module that transforms implicit entity dynamics into explicit causal structures to enhance causal reasoning for crime in VLMs. By combining these two components, our framework advances real-time crime anticipation, achieving improvements in anticipatory tasks: a 70.7% relative improvement in crime classification, a 10.1% in crime detection, and a 3.7% reduction in temporal prediction error.

Recently, multimodal large language models (MLLMs) have emerged as a unified paradigm for language and image generation. Compared with diffusion models, MLLMs possess a much stronger capability for semantic understanding, enabling them to process more complex textual inputs and comprehend richer contextual meanings. However, this enhanced semantic ability may also introduce new and potentially greater safety risks. Taking diffusion models as a reference point, we systematically analyze and compare the safety risks of emerging MLLMs along two dimensions: unsafe content generation and fake image synthesis. Across multiple unsafe generation benchmark datasets, we observe that MLLMs tend to generate more unsafe images than diffusion models. This difference partly arises because diffusion models often fail to interpret abstract prompts, producing corrupted outputs, whereas MLLMs can comprehend these prompts and generate unsafe content. For current advanced fake image detectors, MLLM-generated images are also notably harder to identify. Even when detectors are retrained with MLLMs-specific data, they can still be bypassed by simply providing MLLMs with longer and more descriptive inputs. Our measurements indicate that the emerging safety risks of the cutting-edge generative paradigm, MLLMs, have not been sufficiently recognized, posing new challenges to real-world safety.

Recent progress in video-to-video (V2V) translation has enabled realistic resimulation of embodied AI demonstrations, a capability that allows pretrained robot policies to be transferable to new environments without additional data collection. However, prior works can only operate on a single view at a time, while embodied AI tasks are commonly captured from multiple synchronized cameras to support policy learning. Naively applying single-view models independently to each camera leads to inconsistent appearance across views, and standard transformer architectures do not scale to multi-view settings due to the quadratic cost of cross-view attention. We present VideoWeaver, the first multimodal multi-view V2V translation framework. VideoWeaver is initially trained as a single-view flow-based V2V model. To achieve an extension to the multi-view regime, we propose to ground all views in a shared 4D latent space derived from a feed-forward spatial foundation model, namely, Pi3. This encourages view-consistent appearance even under wide baselines and dynamic camera motion. To scale beyond a fixed number of cameras, we train views at distinct diffusion timesteps, enabling the model to learn both joint and conditional view distributions. This in turn allows autoregressive synthesis of new viewpoints conditioned on existing ones. Experiments show superior or similar performance to the state-of-the-art on the single-view translation benchmarks and, for the first time, physically and stylistically consistent multi-view translations, including challenging egocentric and heterogeneous-camera setups central to world randomization for robot learning.

Referring video object segmentation (RVOS) aims to segment objects in a video described by a natural language expression. However, most existing approaches focus only on the referred object (typically the actor), even when the expression clearly describes an interaction involving multiple objects with distinct roles. In this paper, we introduce Interaction-Aware Referring Video Object Segmentation (InterRVOS), a novel task that focuses on explicit interaction modeling by requiring separate segmentation of actor and target objects.This formulation enables fine-grained understanding of object relationships, as many video events are defined by such interactions rather than individual objects. We present InterRVOS-127K, a large-scale dataset of over 127K automatically annotated expressions with distinct actor-target mask pairs, and propose ReVIOSa, a MLLM-based architecture that introduces interaction-aware special tokens and attention mask loss (AML) to enhance interaction-aware segmentation. We also propose a new evaluation protocol that separately evaluates actor and target segmentation for more accurate role distinction. Comprehensive experiments demonstrate that ReVIOSa outperforms existing baselines on the proposed InterRVOS-127K benchmark, with further analyses validating the necessity and effectiveness of both ReVIOSa and InterRVOS-127K.

Learning-based video quality assessment (VQA) has advanced rapidly, yet progress is increasingly constrained by a disconnect between model design and dataset curation. Model-centric approaches often iterate on fixed benchmarks, while data-centric efforts collect new human labels without systematically targeting the weaknesses of existing VQA models. Here, we describe MDS-VQA, a model-informed data selection mechanism for curating unlabeled videos that are both difficult for the base VQA model and diverse in content. Difficulty is estimated by a failure predictor trained with a ranking objective, and diversity is measured using deep semantic video features, with a greedy procedure balancing the two under a constrained labeling budget. Experiments across multiple VQA datasets and models demonstrate that MDS-VQA identifies diverse, challenging samples that are particularly informative for active fine-tuning. With only a 5% selected subset per target domain, the fine-tuned model improves mean SRCC from 0.651 to 0.722 and achieves the top gMAD rank, indicating strong adaptation and generalization.