CVPR 2026 Papers

3D Gaussian Splatting (3DGS) has emerged as a mainstream representation for 3D scenes, drawing increasing research attention to its understanding, generation, and editing. However, existing studies remain limited to low-level perception, low-quality generation, and low-efficiency editing, lagging far behind their image counterparts in the era of Multimodal Large Language Models (MLLMs). To bridge this gap, we propose MLLMSplat, a novel framework that adapts 2D MLLMs to achieve high-level understanding, high-quality generation, and high-efficiency editing of 3DGS scenes. Specifically, our comprehensive framework consists of three core designs: (1) a 3DGS tokenizer that can be seamlessly integrated into MLLMs in a training-free manner; (2) a 3DGS de-tokenizer that non-intrusively extends the 2D latent diffusion model in MLLMs using a dual rotary positional encoding space, while augmenting it with a jointly trained and sampled 3DGS decoder; and (3) a surrogate task that enhances feed-forward editing capabilities. Extensive experiments demonstrate that MLLMSplat delivers state-of-the-art performance across 3DGS understanding, generation, and editing.

Long video understanding (LVU) remains a core challenge in multimodal learning. Although recent vision-language models (VLMs) have made notable progress, existing benchmarks mainly focus on either fine-grained perception or coarse summarization, offering limited insight into temporal understanding over long contexts. In this work, we define a scene as a coherent segment of a video in which both visual and semantic contexts remain consistent, aligning with human perception. This leads us to a key question: Can current VLMs reason effectively over long, scene-level contexts? To answer this, we introduce a new benchmark, SceneBench, designed to provide scene-level challenges. Our evaluation reveals a sharp drop in accuracy when VLMs attempt to answer scene-level questions, indicating significant forgetting of long-range context. To further validate these findings, we propose Scene Retrieval-Augmented Generation (Scene-RAG), which constructs a dynamic scene memory by retrieving and integrating relevant context across scenes. Scene-RAG improves VLM performance by +2.50%, confirming that current models still struggle with long-context retention. We hope SceneBench will encourage future research toward VLMs with more robust, human-like video comprehension.

Recently proposed pyramidal models decompose the conventional forward and backward diffusion processes into multiple stages operating at varying resolutions. These models handle inputs with higher noise levels at lower resolutions, while less noisy inputs are processed at higher resolutions. This hierarchical approach significantly reduces the computational cost of inference in multi-step denoising models. However, existing open-source pyramidal video models have been trained from scratch and tend to underperform compared to state-of-the-art systems in terms of visual plausibility.In this work, we present a pipeline that converts a pretrained diffusion model into a pyramidal one through low-cost finetuning, achieving this transformation without degradation in quality of output videos.Furthermore, we investigate and compare various strategies for step distillation within pyramidal models, aiming to further enhance the inference efficiency.

Although diffusion transformer (DiT)-based video virtual try-on (VVT) has made significant progress in synthesizing realistic videos, existing methods still struggle to capture fine-grained garment dynamics and preserve background integrity across video frames. They also incur high computational costs due to additional interaction modules introduced into DiTs, while the limited scale and quality of existing public datasets also restrict model generalization and effective training. To address these challenges, we propose a novel framework, KeyTailor, along with a large-scale, high-definition dataset, ViT-HD. The core idea of KeyTailor is a keyframe-driven details injection strategy, motivated by the fact that keyframes inherently contain both foreground dynamics and background consistency. Specifically, KeyTailor adopts an instruction-guided keyframe sampling strategy to filter informative frames from the input video. Subsequently, two tailored keyframe-driven modules--the garment details enhancement module and the collaborative background optimization module--are employed to distill garment dynamics into garment-related latents and to optimize the integrity of background latents, both guided by keyframes. These enriched details are then injected into standard DiT blocks together with pose, mask, and noise latents, enabling efficient and realistic try-on video synthesis. This design ensures consistency without explicitly modifying the DiT architecture, while simultaneously avoiding additional complexity. In addition, our dataset ViT-HD comprises 15,070 high-quality video samples at a resolution of 810 x 1080, covering diverse garments. Extensive experiments demonstrate that KeyTailor outperforms state-of-the-art baselines in terms of garment fidelity and background integrity across both dynamic and static scenarios. The dataset and code will be publicly released.

Vision-Language Models (VLMs) perform well on multimodal benchmarks but lag behind humans and specialized models on visual perception tasks like depth estimation or object counting. Finetuning on one task can unpredictably affect performance on others, making task-specific finetuning challenging. In this paper, we address this challenge through a systematic study of task transferability. We examine how finetuning a VLM on one perception task affects its zero-shot performance on others. We introduce Perfection Gap Factor (PGF), a normalized metric that measures change in performance as a result of task transfer. We utilize PGF to compute Task Transferability, which captures both the breadth and the magnitude of transfer induced by a source task. Using three open-weight VLMs evaluated across 13 perception tasks, we construct a task transfer graph that reveals previously unobserved relationships among perception tasks. Our analysis uncovers patterns of positive and negative transfer, identifies groups of tasks that mutually influence each other, organizes tasks into personas based on their transfer behavior and demonstrates how PGF can guide data selection for more efficient training. These findings highlight both opportunities for positive transfer and risks of negative interference, offering actionable guidance for advancing VLMs.

The synergistic framework of multimodal large language models (MLLMs) and vision foundation models demonstrates exceptional performance in image understanding tasks, yet encounters severe temporal inconsistency challenges in video segmentation scenarios. Existing methods predominantly rely on MLLMs trained on static images to generate per-frame segmentation prompts, neglecting the physical continuity of video motion. This paper posits that performance limitations in video understanding tasks from inadequate constraints on model output behavior. Consequently, we propose a spatiotemporal co-optimization mechanism that achieves temporally consistent video segmentation solely by constraining MLLM output behavior, eliminating the need for large-scale video pretraining or complex architectural modifications. Our method features two complementary mechanisms: a Brownian bridge loss that models object trajectories as endpoint-constrained Gaussian processes to ensure temporal smoothness, and a geometry-aware prompt quality loss that enforces spatial consistency with target structures. Experiments on referring expression video segmentation and reasoning video segmentation tasks demonstrate that our method significantly surpasses state-of-the-art techniques on the Ref-YouTube-VOS, Ref-DAVIS-2017, MeVIS, A2d-Sentences, JHMDB-Sentences and ReVOS benchmarks. This work establishes that explicit modeling of physical world constraints can unlock the full potential of statically trained foundation models in dynamic visual understanding tasks.

Egocentric perception enables humans to experience and understand the world directly from their own point of view. Translating exocentric (third-person) videos into egocentric (first-person) videos opens up new possibilities for immersive understanding but remains highly challenging due to extreme camera pose variations and minimal view overlap. This task requires faithfully preserving visible content while synthesizing unseen regions in a geometrically consistent manner. To achieve this, we present EgoX, a novel framework for generating egocentric videos from a single exocentric input.EgoX leverages the pretrained spatio-temporal knowledge of large-scale video diffusion models through lightweight LoRA adaptation and introduces a unified conditioning strategy that combines exocentric and egocentric priors via width- and channel-wise concatenation.Additionally, a geometry-guided self-attention mechanism selectively attends to spatially relevant regions, ensuring geometric coherence and high visual fidelity.Our approach achieves coherent and realistic egocentric video generation while demonstrating strong scalability and robustness across unseen and in-the-wild videos.

Video snapshot compressive imaging (SCI) offers a promising alternative to high-speed cameras by encoding multiple frames into a single 2D measurement. However, SCI requires algorithms to reconstruct the high-speed video, and as resolution increases, reconstruction becomes computationally expensive and memory-intensive. Much of the resource is wasted on recovering large background regions that contain little useful information, highlighting the need for selective, object-driven reconstruction. Existing object detectors struggle to perform accurately on SCI measurements due to the spatial-temporal aliasing introduced by coded exposure. To address this challenge, we propose DetectSCI, the first framework enabling object-guided region-of-interest (ROI) reconstruction for high-resolution SCI. The inside detector comprises two key components: an encoder built from weight-sharing Mamba-Implicit Modules (MIM) for progressive feature refinement, and a Frequency Mamba (FM) module dedicated to frequency-aware query selection. MIM enhances features via multi-scale dilated convolutions and implicit representations, while FM restores discriminative details by decomposing and reweighting frequency bands. Experiments on the SportsMOT dataset show that DetectSCI achieves 80.9 Average Precision (AP), surpassing the best CNN-based detector by at least 2.8 AP and the best Transformer-based detector by at least 4.1 AP, while maintaining comparable efficiency. Code will be released.

Existing methods for human motion control in video generation typically rely on either 2D poses or explicit 3D parametric models (e.g., SMPL) as control signals. However, 2D poses rigidly bind motion to the driving viewpoint, precluding novel-view synthesis. Explicit 3D models, though structurally informative, suffer from inherent inaccuracies (e.g., depth ambiguity and inaccurate dynamics) which, when used as a strong constraint, override the powerful intrinsic 3D awareness of large-scale video generators. In this work, we revisit motion control from a 3D-aware perspective, advocating for an implicit, view-agnostic motion representation that naturally aligns with the generator's spatial priors rather than depending on externally reconstructed constraints. We introduce 3DiMo, which jointly trains a motion encoder with a pretrained video generator to distill driving frames into compact, view-agnostic motion tokens, injected semantically via cross-attention. To foster 3D awareness, we train with view-rich supervision--single-view, multi-view, and moving-camera videos--forcing motion consistency across diverse viewpoints. Additionally, we use auxiliary geometric supervision that leverages SMPL only for early initialization and is annealed to zero, enabling the model to transition from external 3D guidance to learning genuine 3D spatial motion understanding from the data and the generator's priors. Experiments confirm that 3DiMo faithfully reproduces driving motions with flexible, text-driven camera control, significantly surpassing existing methods in both motion fidelity and visual quality.

Reinforcement learning (RL) has emerged as an effective approach for improving video reasoning in multimodal large language models (MLLMs). However, existing methods remain inefficient for two reasons. First, training data are typically organized by task formats rather than underlying reasoning abilities, creating a mismatch that encourages models to learn task-specific patterns instead of transferable capabilities. As a result, improving ability generalization requires broad coverage over many ability-task combinations, making RL training costly. Second, prior work often compensates for this inefficiency with increasingly complex algorithmic designs, such as specialized temporal architectures or multi-objective reward frameworks, which further complicate training. To address these issues, we present VAST, a cognitive taxonomy and data organization framework that structures video understanding into three layers: Perception, Reasoning, and Cognition. Based on this taxonomy, we construct VAST-15K for training and VAST-Bench for evaluation. We further introduce VideoVAST, a reinforcement learning framework that uses consistency rewards to encourage alignment between reasoning traces and final answers, without architectural modifications. Experiments show that VideoVAST achieves 66.3% accuracy on MVBench and 57.4% on VAST-Bench, compared with 62.7% and 54.3% for Video-R1, while using approximately 72% fewer GPU hours and 96% fewer training samples under the same training settings. Code and resources are available at https://zhongan-wang.github.io/VAST.

Existing online video segmentation models typically combine a per-frame segmenter with complex specialized tracking modules. While effective, these modules introduce significant architectural complexity and computational overhead. Recent studies suggest that plain Vision Transformer (ViT) encoders, when scaled with sufficient capacity and large-scale pre-training, can conduct accurate image segmentation without requiring specialized modules. Motivated by this observation, we propose the Video Encoder-only Mask Transformer (VidEoMT), a simple encoder-only video segmentation model that eliminates the need for dedicated tracking modules. To enable temporal modeling in an encoder-only ViT, VidEoMT introduces a lightweight query propagation mechanism that carries information across frames by reusing queries from the previous frame. To balance this with adaptability to new content, it employs a query fusion strategy that combines the propagated queries with a set of temporally-agnostic learned queries. As a result, VidEoMT attains the benefits of a tracker without added complexity, achieving competitive accuracy while being 5x-10x faster, running at up to 160 FPS with a ViT-L backbone. Code: \href https://www.tue-mps.org/videomt/ https://www.tue-mps.org/videomt/ .

Modern cameras, such as smartphone cameras and DSLRs, are equipped with gyro sensors that measure motion of the camera. While the motion information is valuable for deblurring, gyro-based deblurring has not been widely studied, particularly for video. A few gyro-based video deblurring methods have been proposed, but they exhibit inherent limitations. First, gyro sensors capture only rotational motion, leading these methods to ignore translational motion. Second, their dependence on simplified blur models and deconvolution-based solutions restricts overall performance. To address these limitations, we introduce GyroDVD, the first learning-based framework for gyro-based video deblurring. We propose a novel blur kernel construction scheme that jointly accounts for rotational and translational motion. A video deblurring network then restores sharp videos by exploiting the constructed kernels together with the video frames. For training and evaluation, we introduce the GyroVD dataset, a large-scale and realistic dataset specifically designed for gyro-based deblurring. Extensive experiments demonstrate that our method significantly outperforms prior gyro-based image and video deblurring methods.

Video Unsupervised Domain Adaptation (VUDA) poses a significant challenge in action recognition, requiring the adaptation of a model from a labeled source domain to an unlabeled target domain. Despite recent advances, existing VUDA methods often fall short of fully supervised performance, a key reason being the prevalence of static and uninformative backgrounds that exacerbate domain shifts. Additionally, prior approaches largely overlook computational efficiency, limiting real-world adoption. To address these issues, we propose Learnable Motion-Focused Tokenization (LMFT) for VUDA. LMFT tokenizes video frames into patch tokens and learns to discard low-motion, redundant tokens, primarily corresponding to background regions, while retaining motion-rich, action-relevant tokens for adaptation. Extensive experiments on three standard VUDA benchmarks across 21 domain adaptation settings show that our VUDA framework with LMFT achieves state-of-the-art performance while significantly reducing computational overhead. LMFT thus enables VUDA that is both effective and computationally efficient.

Video amodal completion (VAC) aims to mimic the human brain's ability to implicitly perceive the complete appearance of partially occluded objects, thereby facilitating recognition and understanding. Existing VAC methods finetune video generation models on custom datasets, yet these datasets often have unrealistic distributions and small scales due to the challenges of collecting real amodal data, and thus limit their performance and generalization. To address this, we propose a training-free framework by utilizing pre-trained image inpainting models for VAC and introducing in-context (IC) learning to enhance inter-frame consistency. However, despite the satisfactory performance of DiT-based IC Learning in generation tasks, task-agnostic global attention tends to utilize irrelevant contextual information, resulting in completion failures when applied to amodal completion. Additionally, IC Learning faces a cold-start problem with the exemplar construction. To this end, we propose a consistent video amodal completion with rectified in-context exemplar guidance. Specifically, we introduce rectified exemplar-guided completion by adjusting the attention weights of exemplar relative to the target images for consistent completion, and adopt a dual-frame calibrated exemplar rectification to tackle the cold-start issue. Quantitative and qualitative experiments demonstrate that our method outperforms SOTAs, especially in terms of generalization and robustness on uncommon data and under severe occlusion. Codes are available at https://github.com/xxyCSCV/IC-Amodal.

Video-language models (VLMs) learn to reason about the dynamic visual world through natural language. We introduce a suite of open datasets, benchmarks, and recipes for scalable oversight that enable precise video captioning. First, we define a structured specification for describing subjects, scenes, motion, spatial, and camera dynamics, grounded by hundreds of carefully defined visual primitives developed with professional video creators such as filmmakers. Next, to curate high-quality captions, we introduce a critique-based human-AI (CHAI) oversight framework, where trained human experts provide correctional critiques to revise model-generated pre-captions into improved post-captions. This division of labor improves annotation accuracy and efficiency by offloading text generation to models, allowing humans to better focus on verification. Additionally, these critiques and preferences between pre- and post-captions provide rich supervision for improving open-source models (Qwen3-VL) on caption generation, reward modeling, and critique generation through SFT, DPO, and inference-time scaling. Our ablations show that critique quality in precision, recall, and constructiveness, ensured by our oversight framework, directly governs downstream performance. With modest expert supervision, the resulting model outperforms even closed-source models such as Gemini-3.1-Pro. Finally, we apply our approach to re-caption large-scale professional videos (e.g., films, commercials, games) and fine-tune video generation models such as Wan to better follow detailed prompts of up to 400 words, achieving finer control over cinematography including camera motion, angle, lens, focus, point of view, and framing. Overall, our results show that precise specification and human-AI oversight are key to achieving professional-level video understanding and generation.

Recent advances in self-supervised learning (SSL) for point clouds have substantially improved 3D scene understanding without human annotations. Existing approaches emphasize semantic awareness by enforcing feature consistency across augmented views or by masked scene modeling. However, the resulting representations transfer poorly to instance localization, and often require full finetuning for strong performance. Instance awareness is a fundamental component of 3D perception, thus bridging this gap is crucial for progressing toward true 3D foundation models that support all downstream tasks on 3D data. In this work, we introduce PointINS, an instance-oriented self-supervised framework that enriches point cloud representations through geometry-aware learning.PointINS employs an orthogonal offset branch to jointly learn high-level semantic understanding and geometric reasoning, yielding instance awareness. We identify two consistent properties essential for robust instance localization and formulate them as complementary regularization strategies, Offset Distribution Regularization (ODR), which aligns predicted offsets with empirically observed geometric priors, and Spatial Clustering Regularization (SCR), which enforces local coherence by regularizing offsets with pseudo-instance masks. Through extensive experiments across five datasets, PointINS achieves on average +3.5% mAP improvement for indoor instance segmentation and +4.1% PQ gain for outdoor panoptic segmentation, paving the way for scalable 3D foundation models.

Video Object-Centric Learning seeks to decompose raw videos into a small set of object slots, but existing slot-attention models often suffer from severe over-fragmentation. This is because the model is implicitly encouraged to occupy all slots to minimize the reconstruction objective, thereby representing a single object with multiple redundant slots. We tackle this limitation with a reconstruction-guided slot curriculum (SlotCurri). Training starts with only a few coarse slots and progressively allocates new slots where reconstruction error remains high, thus expanding capacity only where it is needed and preventing fragmentation from the outset. Yet, during slot expansion, meaningful sub-parts can emerge only if coarse-level semantics are already well separated; however, with a small initial slot budget and an MSE objective, semantic boundaries remain blurry. Therefore, we augment MSE with a structure-aware loss that preserves local contrast and edge information to encourage each slot to sharpen its semantic boundaries. Lastly, we propose a cyclic inference that rolls slots forward and then backward through the frame sequence, producing temporally consistent object representations even in the earliest frames. All combined, SlotCurri addresses object over-fragmentation by allocating representational capacity where reconstruction fails, further enhanced by structural cues and cyclic inference. Notable FG-ARI gains of +6.8 on YouTube-VIS and +8.3 on MOVi-C validate the effectiveness of SlotCurri. Our code is available at github.com/wjun0830/SlotCurri.

A central challenge in visual intelligence is learning the physical structure of scenes from raw videos: how regions form objects and the laws that govern their interactions. Solving these tasks requires world models capable of inferring distributional states of the world from partial observations -- capabilities that current architectures do not provide. We introduce a new class of probabilistic world models that support estimation of the probability of any visual variable, such as appearance and dynamics, conditioned on any other variables. Here, we identify that these models can be trained efficiently with autoregressive sequence modeling, yielding world models from which rich object understanding emerges. First, we demonstrate that our model captures the physical laws governing how objects move by generating multiple plausible future states of the world through sequential inference. Then, by analyzing motion correlations across these futures, we extract objects and articulated object subparts. Having discovered these objects, we show that our world model can manipulate them in 3D. Finally, we demonstrate how physical relationships between objects can be computed from the world model, enabling applications such as Visual Jenga.

The "Thinking with Text" and "Thinking with Images" paradigms significantly improve the reasoning abilities of large language models (LLMs) and Vision-Language Models (VLMs). However, these paradigms have inherent limitations. (1) Images capture only single moments and fail to represent dynamic processes or continuous changes, and (2) The separation of text and vision as distinct modalities, which hinders unified multimodal understanding and generation. Therefore, we propose "Thinking with Video", a new paradigm that leverages video generation models such as Sora-2 to use video frames as a unified medium for multimodal reasoning. To support this exploration, we developed the Video Thinking Benchmark (VideoThinkBench), which covers both vision-centric tasks (e.g., Eyeballing Puzzles) and text-centric tasks (e.g., GSM8K and MMMU). Our evaluation on VideoThinkBench establishes Sora-2 as a capable reasoner. On vision-centric tasks, Sora-2 is comparable to state-of-the-art (SOTA) VLMs, and even surpasses GPT-5 by 10% on eyeballing puzzles. On text-centric tasks, Sora-2 achieves 92% accuracy on MATH, and 69.2% accuracy on MMMU. Furthermore, we systematically analyze the source of these abilities. We also find that self-consistency and in-context learning can improve Sora-2's performance. In summary, our findings show that the video generation model is the potential unified multimodal understanding and generation model, positioning "Thinking with Video" as a potential unified multimodal reasoning paradigm.