CVPR 2026 论文列表

Diffusion models have shown impressive performance in generative tasks, demonstrating their ability to capture detailed structural and semantic information. Recently, these capabilities have been extended to visual understanding, with studies employing diffusion models as the backbone for various perception tasks. However, existing diffusion-based perception models are generally restricted to a single task or a fixed set of predefined tasks, lacking an efficient mechanism to generalize to novel tasks. To overcome this limitation, we propose a unified DiT-based perception framework called UniPercept, which introduces a novel foundation-adapter paradigm for general visual perception. In this framework, a shared diffusion-based foundation model is trained to capture common and generalizable visual knowledge across diverse perception tasks, with task-specific adapters integrated for each individual task. Leveraging its superior generalization ability, the foundation model can be efficiently adapted to novel domains through lightweight adapters, requiring as few as **1,000** training samples and less than **1%** of trainable parameters. Furthermore, UniPercept demonstrates strong performance across various perception tasks, outperforming state-of-the-art generalist models in most cases and achieving accuracy comparable to specialist models. Project page: https://VIPL-GENUN.github.io/Project-UniPercept.

Generating high-fidelity audio that is both semantically meaningful and temporally synchronized with silent videos remains a challenging problem in video-to-audio generation. Existing approaches often fail to capture fine-grained temporal correspondence between visual events and audio dynamics, leading to unrealistic or desynchronized outputs. To address these limitations, we propose VisioSonic, a Video-Aligned Sound generation framework that unifies flow-matching diffusion and preference-guided alignment. VisioSonic introduces a multimodal conditioning module that jointly leverages video frames and textual cues to provide semantic and frame-level temporal guidance. A co-attention diffusion transformer efficiently fuses visual and audio representations, enabling content-aware sound synthesis with minimal computation costs. To further enhance alignment beyond supervised training, we introduce Semantic-Temporal Alignment Ranked Direct Preference Optimization (STAR-DPO), a novel preference-learning paradigm that automatically generates audio candidates, ranks them based on both semantic and temporal alignment, and subsequently fine-tunes the diffusion model using the derived preference pairs. Extensive experiments on various benchmarks demonstrate that VisioSonic achieves state-of-the-art audio-video synchronization and audio fidelity while using the fewest trainable parameters among competing approaches. Project page: https://kaiw7.github.io/VisioSonic/

Generalizing video matting models to real-world videos remains a significant challenge due to the scarcity of labeled data. To address this, we present Video Mask-to-Matte Model VideoMaMa that converts coarse segmentation masks into pixel accurate alpha mattes, by leveraging pretrained video diffusion models. VideoMaMa demonstrates strong zero-shot generalization to real-world footage, even though it is trained solely on synthetic data. Building on this capability, we develop a scalable pseudo-labeling pipeline for large-scale video matting and construct the Matting Anything in Video MA-V dataset, which offers high-quality matting annotations for more than 50K real-world videos spanning diverse scenes and motions. To validate the effectiveness of this dataset, we fine-tune the SAM2 model on MAV to obtain SAM2-Matte, which outperforms the same model trained on existing matting datasets in terms of robustness on in-the-wild videos.These findings emphasize the importance of large-scale pseudo-labeled video matting and showcase how generative priors and accessible segmentation cues can drive scalable progress in video matting research.

Diffusion models have recently advanced video restoration, but applying them to real-world and AIGC-generated video super-resolution (VSR) remains challenging due to high latency, prohibitive computation, and poor generalization to ultra-high resolutions. Our goal in this work is to make diffusion-based VSR practical by achieving efficiency, scalability, and near real-time performance. To this end, we propose FlashVSR, the first diffusion-based one-step streaming framework for efficient video super-resolution. FlashVSR runs at 17 FPS for 768x1408 videos on a single A100 GPU by combining three complementary innovations: (i) a train-friendly three-stage distillation pipeline that enables streaming super-resolution, (ii) locality-constrained sparse attention that reduces redundant computation while bridging the train--test resolution gap, and (iii) a tiny conditional decoder that accelerates reconstruction without sacrificing quality. To support large-scale training, we also construct VSR-120K, a new dataset containing 120 K videos and 180 K images. Extensive experiments demonstrate that FlashVSR scales reliably to ultra-high resolutions and achieves state-of-the-art performance with up to 12x speed-up over prior one-step diffusion-based VSR models. We will release the code, pretrained models, and dataset to foster future research in efficient diffusion-based video super-resolution.

Single-image-to-3D shape generation has seen remarkable progress, driven by latent diffusion models trained on the compressed latent space of 3D VAEs. However, the task remains intrinsically ill-posed: recovering complete 3D geometry--especially occluded surfaces--from a single view is inherently ambiguous. Existing VecSet-based approaches further exacerbate this challenge by treating shape tokens as an unordered set without explicit positional encoding. This design forces diffusion models to simultaneously learn visible correspondences from the input image and hallucinate invisible geometry within a large, permutation-invariant token space, where the lack of structure significantly hinders training efficiency and convergence stability.To address this, we propose Visibility Learning, a training paradigm that injects visibility structure and positional inductive bias into the image-to-3D pipeline. Our method comprises two synergistic components: (1) Visibility Grouping (VG), which explicitly partitions VecSet tokens into visible and invisible subsets by exploiting the spatial locality of VecSet VAE decoders; and (2) Visibility-Aware Positional Encoding (VAPE), which assigns shared positional embeddings to image tokens and visible shape tokens to amplify their correspondence, while using distinct encodings for invisible tokens to guide hallucination. By explicitly disentangling visible reconstruction from invisible hallucination, our approach shrinks the effective hypothesis space and provides clear structural guidance for diffusion models. Extensive experiments demonstrate that Visibility Learning accelerates training convergence by up to 4.4xwhile achieving superior generation quality compared to strong VecSet-based baselines.

Masked auto-regressive diffusion models (MAR) benefit from the expressive modeling ability of diffusion models and the flexibility of masked auto-regressive ordering. However, vanilla MAR suffers from slow inference due to its hierarchical inference mechanism: an outer AR unmasking loop and an inner diffusion denoising chain. Such decoupled structure not only harm the generation efficiency but also hinder the practical use of MAR for reinforcement learning (RL), an increasingly critical paradigm for generative model post-training.To address this fundamental issue, we introduce MARVAL (Masked Auto-regressive Variational Acceleration), a distillation-based framework that compresses the diffusion chain into a single AR generation step while preserving the flexible auto-regressive unmasking order. Such a distillation with MARVAL not only yields substantial inference acceleration but, crucially, makes RL post-training with verifiable rewards practical, resulting in scalable yet human-preferred fast generative models. Our contributions are twofold: (1) a novel score-based variational objective for distilling masked auto-regressive diffusion models into a single generation step without sacrificing sample quality; and (2) an efficient RL framework for masked auto-regressive models via MARVAL-RL. On ImageNet 256x256, MARVAL-Huge achieves an FID of 2.00 with more than 30 times speedup compared with MAR-diffusion, and MARVAL-RL yields consistent improvements in CLIP and image-reward scores on ImageNet datasets with entity names. In conclusion, MARVAL demonstrates the first practical path to distillation and RL of masked auto-regressive diffusion models, enabling fast sampling and better preference alignments. Project page: https://ai4imaginglab.github.io/MARVAL-RL-Page/

Recent progress in 3D generation has been driven largely by models conditioned on images or text, while readily available 3D priors are still underused. In many real-world scenarios, the visible-region point cloud are easy to obtain--from active sensors such as LiDAR or from feed-forward predictors like VGGT--offering explicit geometric constraints that current methods fail to exploit. In this work, we introduce Points-to-3D, a diffusion-based framework that leverages point cloud priors for geometry-controllable 3D asset and scene generation. Built on a latent 3D diffusion model TRELLIS, Points-to-3D first replaces pure-noise sparse structure latent initialization with a point cloud priors tailored input formulation. A structure inpainting network, trained within the TRELLIS framework on task-specific data designed to learn global structural inpainting, is then used for inference with a staged sampling strategy (structural inpainting followed by boundary refinement), completing the global geometry while preserving the visible regions of the input priors.In practice, Points-to-3D can take either accurate point-cloud priors or VGGT-estimated point clouds from single images as input.Experiments on both single-object and multi-object scenarios consistently demonstrate superior performance over state-of-the-art baselines in terms of rendering quality and geometric fidelity, highlighting the effectiveness of explicitly embedding point-cloud priors for achieving more accurate and structurally controllable 3D generation.

Diffusion Multi-modal Large Language Models (dMLLMs) have recently emerged as a novel architecture unifying image generation and understanding. However, developing effective and efficient Test-Time Scaling (TTS) methods to unlock their full generative potential remains an underexplored challenge. To address this, we propose dMLLM-TTS, a novel framework operating on two complementary scaling axes: (1) trajectory exploration scaling to enhance the diversity of generated hypotheses, and (2) iterative refinement scaling for stable generation. Conventional TTS approaches typically perform linear search across these two dimensions, incurring substantial computational costs of O(NT) and requiring an external verifier for best-of-N selection. To overcome these limitations, we propose two innovations. First, we design an efficient hierarchical search algorithm with O(N+T) complexity that adaptively expands and prunes sampling trajectories. Second, we introduce a self-verified feedback mechanism that leverages the dMLLMs' intrinsic image understanding capabilities to assess text-image alignment, eliminating the need for external verifier. Extensive experiments on the GenEval benchmark across three representative dMLLMs (e.g., Lumina-DiMOO, MMaDA, Muddit) show that our framework substantially improves generation quality while achieving up to 6x greater efficiency than linear search.

Motion forecasting predicts where points will move in the future, while motion tracking predicts where they are in the present. Despite these similarities, existing approaches to the two problems are quite different. In this paper, we propose a unified model that can address both tasks. We train a causal, video-conditioned flow matching model to predict point positions. The resulting model can easily toggle between point tracking to forecasting by changing its visual signal. Despite our model's simplicity, we find that it outperforms prior work in point forecasting and obtains performance that is competitive with the state-of-the-art on the TAP-Vid benchmark.

Recent text-to-image (T2I) diffusion models have achieved impressive progress in generating high-fidelity images, yet they often fail to faithfully follow complex user prompts, especially in attribute binding, negation, and compositional reasoning. To address this limitation, we propose PromptEnhancer, a universal prompt rewriting framework that improves prompt interpretability for any pre-trained T2I model. PromptEnhancer is trained through a multi-stage pipeline. We first perform supervised fine-tuning on chain-of-thought-style rewriting data to endow the rewriter with structured prompt analysis and rewriting capabilities. We then introduce AlignEvaluator, a task-specific reward model that provides explicit and fine-grained feedback based on a taxonomy of common T2I failure modes, and further optimize the rewriter with reinforcement learning. This design enables the rewriter to produce prompts that are more precise, semantically complete, and easier for T2I models to follow. To support evaluation, we also construct a comprehensive human-aligned benchmark covering diverse semantic and compositional challenges. Extensive experiments show that PromptEnhancer consistently improves image-text alignment across multiple T2I models and significantly enhances prompt fidelity on challenging cases.

Image protection against unauthorized diffusion-based editing has achieved encouraging progress. However, existing methods face two critical limitations: (1) They only disturb the denoising direction at local step, resulting in generated images still retaining original or edited semantics. (2) Their optimization rely heavily on model-specific gradient, limiting transferable protection across different models and tasks. To address these challenges, we propose a Universal Defense (UniDef) framework for protection against unauthorized image manipulation. Specifically, we first discover that different variants of diffusion models tend to pursue a consistent distribution objective during complete denoising process. Based on this discovery, we design Consistent Distribution Deviation strategy to perturb the diffusion direction at the global denoising, thereby disrupting the overall image semantics. Furthermore, to mitigate model dependency, we devise a Finite Difference-based Jacobian Estimation module to approximate the global gradient in a model-agnostic manner, thus ensuring more transferable protection. Benefiting from the above designs, our method yields generated images no longer preserve the image semantic while possessing excellent generalization. Extensive experiments demonstrate that our UniDef not only outperforms existing methods, but also exhibits universal protection across diverse models and tasks.

Text-to-image (T2I) diffusion models such as SDXL and FLUX have achieved impressive photorealism, yet small-scale distortions remain pervasive in limbs, face, text and so on. Existing refinement approaches either perform costly iterative re-generation or rely on vision-language models (VLMs) with weak spatial grounding, leading to semantic drift and unreliable local edits. To close this gap, we propose **Agentic Retoucher**, a hierarchical decision-driven framework that reformulates post-generation correction as a human-like *perception-reasoning-action* loop.Specifically, we design (1) a **perception agent** that learns contextual saliency for fine-grained distortion localization under text-image consistency cues, (2) a **reasoning agent** that performs human-aligned inferential diagnosis via progressive preference alignment, and (3) an **action agent** that adaptively plans localized inpainting guided by user preference. This design integrates perceptual evidence, linguistic reasoning, and controllable correction into a unified, self-corrective decision process. To enable fine-grained supervision and quantitative evaluation, we further construct **GenBlemish-27K**, a dataset of 6K T2I images with 27K annotated artifact regions across 12 categories.Extensive experiments demonstrate that Agentic Retoucher consistently outperforms state-of-the-art methods in perceptual quality, distortion localization and human preference alignment, establishing a new paradigm for self-corrective and perceptually reliable T2I generation.

Concept erasure in text-to-image diffusion models seeks to remove undesired concepts while preserving overall generative capability. Localized erasure methods aim to restrict edits to the spatial region occupied by the target concept. However, we observe that suppressing a concept can unintentionally weaken semantically related neighbor concepts, reducing fidelity in fine-grained domains. We propose Neighbor-Aware Localized Concept Erasure (NLCE), a training-free framework designed to better preserve neighboring concepts while removing target concepts. It operates in three stages: (1) a spectrally-weighted embedding modulation that attenuates target concept directions while stabilizing neighbor concept representations, (2) an attention-guided spatial gate that identifies regions exhibiting residual concept activation, and (3) a spatially-gated hard erasure that eliminates remaining traces only where necessary. This neighbor-aware pipeline enables localized concept removal while maintaining the surrounding concept neighborhood structure. Experiments on fine-grained datasets (Oxford Flowers, Stanford Dogs) show that our method effectively removes target concepts while better preserving closely related categories. Additional results on celebrity identity, explicit content and artistic style demonstrate robustness and generalization to broader erasure scenarios.

Acceleration methods for diffusion models (e.g., token merging or downsampling) typically optimize for synthesis quality under reduced compute, yet they often ignore the model's latent discriminative capacity. We revisit token compression with a joint objective and present **BiGain**, a training-free, plug-and-play framework that preserves generation quality while markedly improving classification in accelerated diffusion models. Our key insight is frequency separation: mapping feature-space signals into a frequency-aware representation disentangles fine detail from global semantics, enabling compression that respects both generative fidelity and discriminative utility. BiGain reflects this principle with two frequency-aware operators: (1) *Laplacian-gated token merging*, which encourages merges among spectrally smooth tokens while discouraging merges of high-contrast tokens, thereby retaining edges and textures; and (2) *Interpolate-Extrapolate KV Downsampling*, which downsamples keys/values via a controllable interextrapolation between nearest and average pooling while keeping queries intact, thereby conserving attention precision without retraining. Across DiT- and U-Net-based backbones and multiple datasets of ImageNet-1K, ImageNet-100, Oxford-IIIT Pets, and COCO-2017, our proposed operators consistently improve the speed-accuracy trade-off for diffusion-based classification, while maintaining, sometimes even enhancing generation quality under comparable acceleration. For instance, on ImageNet-1K, with 70% token merging ratio on Stable Diffusion 2.0, BiGain increases classification accuracy by 7.15% while also improving FID for generation by 0.34 (1.85%). Our comprehensive analyses indicate that balanced spectral retention, preserving high-frequency detail alongside low/mid-frequency semantic content is a reliable design rule for token compression in diffusion models. To our knowledge, BiGain is the first framework to jointly study and advance both generation and classification under accelerated diffusion, supporting lower-cost deployment of dual-purpose generative systems.

Existing approaches to 3D semantic urban scene generation predominantly rely on voxel-based representations, which are bound by fixed resolution, challenging to edit, and memory-intensive in their dense form. In contrast, we advocate for a primitive-based paradigm where urban scenes are represented using compact, semantically meaningful 3D elements that are easy to manipulate and compose. To this end, we introduce PrITTI, a latent diffusion model that leverages vectorized object primitives and rasterized ground surfaces for generating diverse, controllable, and editable 3D semantic urban scenes. This hybrid representation yields a structured latent space that facilitates object- and ground-level manipulation. Experiments on KITTI-360 show that primitive-based representations unlock the full capabilities of diffusion transformers, achieving state-of-the-art 3D scene generation quality with lower memory requirements, faster inference, and greater editability than voxel-based methods. Beyond generation, PrITTI supports a range of downstream applications, including scene editing, inpainting, outpainting, and photo-realistic street-view synthesis.

Despite recent progress, reinforcement learning (RL)-based fine-tuning of diffusion models often struggles with generalization, composability, and robustness against reward hacking. Recent studies have explored prompt refinement as a modular alternative, but most adopt a feed-forward approach that applies a single refined prompt throughout the entire sampling trajectory, thereby failing to fully leverage the sequential nature of reinforcement learning. To address this, we introduce PromptLoop, a plug-and-play RL framework that incorporates latent feedback into step-wise prompt refinement. Rather than modifying diffusion model weights, a multimodal large language model (MLLM) is trained with RL to iteratively update prompts based on intermediate latent states of diffusion models. This design achieves a structural analogy to the Diffusion RL approach, while retaining the flexibility and generality of prompt-based alignment. Extensive experiments across diverse reward functions and diffusion backbones demonstrate that PromptLoop (i) achieves effective reward optimization, (ii) generalizes seamlessly to unseen models, (iii) composes orthogonally with existing alignment methods, and (iv) mitigates over-optimization and reward hacking while introducing only a practically negligible inference overhead.

Point cloud completion is crucial for robust 3D perception but remains challenging. Coarse-to-fine methods can lead to unconstrained local guesses in the absence of key structures, whereas diffusion-based approaches may introduce geometric inconsistencies. To overcome these limitations, we present TouchDream, a novel framework that leverages a diffusion model to `dream' of tactile sensing on object surfaces, which reformulates the sensing process as a learnable generative modeling task. Unlike visual cues, tactile data provides rich local geometry that can be directly converted into 3D space for point fusion, offering a powerful guide for detail-aware completion. Specifically, our approach generate compact tactile latent representations conditioned on coarse points and sampled touch poses. These generated touches are then used to optimize the coarse geometry. Extensive experiments show that our TouchDream model achieves the state-of-the-art performance.

We present DAWN (Diffusion is All We Need for robot control), a unified diffusion-based framework for language-conditioned robotic manipulation that bridges high-level motion intent and low-level robot action via structured pixel motion representation. In DAWN, both the high-level and low-level controllers are modeled as diffusion processes, yielding a fully trainable, end-to-end system with interpretable intermediate motion abstractions. DAWN achieves state-of-the-art results on the challenging CALVIN benchmark, demonstrating strong multi-task performance, and further validates its effectiveness on MetaWorld. Despite the substantial domain gap between simulation and reality and limited real-world data, we demonstrate reliable real-world transfer with only minimal finetuning, illustrating the practical viability of diffusion-based motion abstractions for robotic control. Our results show the effectiveness of combining diffusion modeling with motion-centric representations as a strong baseline for scalable and robust robot learning.

Diffusion-based generative models have achieved remarkable success in real-world image super-resolution (SR). With tiled diffusion techniques, these models can produce high-resolution images that exceed their native-supported resolution. However, the quality of such high-resolution (e.g 2048^2) outputs often remains extremely poor, primarily due to two factors we consider: the image upsampling ratio (e.g x8) exceeding the model's native-supported upsampling ratio (e.g x4), and the model's native-supported resolution. In practice, training a native high-resolution model requires larger architectures, which incur significant computational overhead and GPU memory costs, making it hard on limited-resource equipment. Thus, we present TUDSR, a Twice Upsampling-Diffusion framework for higher SR. The TUDSR framework mainly consists of two stages: the first involves training at R-resolution, and the second introduces a looped chunk-based training strategy at NR-resolution. Each stage adapts a one-step GAN architecture comprising a generator and a discriminator. Based on SD2.1-base, we develop TUDSR-S, which achieves state-of-the-art performance across multiple benchmarks. Extensive experiments further demonstrate that TUDSR-S generates high-quality images at the resolutions of 1024^2 and even 2048^2, significantly outperforming existing approaches. Code is available at https://github.com/wuer5/TUDSR.