CVPR 2026 论文列表

While recent Transformer and Mamba architectures have advanced point cloud representation learning, they are typically developed for single-task or single-domain settings. Directly applying them to multi-task domain generalization (DG) leads to degraded performance. Transformers effectively model global dependencies but suffer from quadratic attention cost and lack explicit structural ordering, whereas Mamba offers linear-time recurrence yet often depends on coordinate-driven serialization, which is sensitive to viewpoint changes and missing regions, causing structural drift and unstable sequential modeling. In this paper, we propose Structure-Aware Domain Generalization (SADG), a Mamba-based In-Context Learning framework that preserves structural hierarchy across domains and tasks. We design structure-aware serialization (SAS) that generates transformation-invariant sequences using centroid-based topology and geodesic curvature continuity. We further devise hierarchical domain-aware modeling (HDM) that stabilizes cross-domain reasoning by consolidating intra-domain structure and fusing inter-domain relations. At test time, we introduce a lightweight spectral graph alignment (SGA) that shifts target features toward source prototypes in the spectral domain without updating model parameters, ensuring structure-preserving test-time feature shifting. In addition, we introduce MP3DObject, a real-scan object dataset for multi-task DG evaluation. Comprehensive experiments demonstrate that the proposed approach improves structural fidelity and consistently outperforms state-of-the-art methods across multiple tasks including reconstruction, denoising, and registration.

World models provide a powerful framework for simulating environment dynamics conditioned on actions or instructions, enabling downstream tasks such as action planning or policy learning.Recent approaches leverage world models as learned simulators, but its application to decision-time planning remains computationally prohibitive for real-time control.A key bottleneck lies in latent representations: conventional tokenizers encode each observation into hundreds of tokens, making planning both slow and resource-intensive.To address this, we propose CompACT, a discrete tokenizer that compresses each observation into as few as 8 tokens, drastically reducing computational cost while preserving essential information for planning.An action-conditioned world model that occupies CompACT tokenizer achieves competitive planning performance with orders-of-magnitude faster planning, offering a practical step toward real-world deployment of world models.

Counterfactual explanations (CFEs) are minimal and semantically meaningful modifications of the input of a model that alter the model predictions. They highlight the decisive features the model relies on, providing contrastive interpretations for classifiers. State-of-the-art visual counterfactual explanation methods have primarily focused on interpreting image classifiers, leaving the domain of video models relatively underexplored. For the video CFEs to be useful, they have to be physically plausible, temporally coherent, and exhibit smooth motion trajectories. Existing CFE image-based methods, designed to explain image classifiers, lack the capacity to generate temporally coherent, smooth and physically plausible video CFEs. To address this, we propose Back To The Feature (BTTF), an optimization framework that generates video CFEs. Our method introduces two novel features, 1) an optimization scheme to retrieve the initial latent noise conditioned by the first frame of the input video, 2) a two-stage optimization strategy to enable the search for counterfactual videos in the vicinity of the input video. Both optimization processes are guided solely by the target classifier, ensuring the explanation is faithful. To accelerate convergence, we also introduce a progressive optimization strategy that incrementally increases the number of denoising steps. Extensive experiments on video datasets such as Shape-Moving (motion classification), MEAD (emotion classification), and NTU RGB+D (action classification) show that our BTTF effectively generates valid, visually similar and realistic counterfactual videos that provide concrete insights into the classifier's decision-making mechanism.

Photorealistic style transfer aims to match the color and tone of an input image to that of a style target while preserving the content and details of the original scene. Although existing large image models can facilitate these kinds of appearance edits, their high computational demands, potential for hallucinations, and limited user control make them unsuitable for high-resolution, real-time workflows. We introduce Hist2Style, a bilateral-grid formulation for fast, edge-aware stylization that preserves visual fidelity by constraining operations to locally affine transforms in bilateral space. Our model distills a large image editing model into a lightweight network by training on a large supervised corpus generated with language and vision-language models, targeting spatially varying color edits. The network conditions on a histogram-based embedding of the style target to provide an interpretable interface for adjusting the output style by modifying the target color distribution. Overall, Hist2Style maintains content structure by construction, avoids hallucinations, and supports real-time, high-resolution photorealistic stylization with interactive user-controllable color and tone adjustments. Our project page is available at \href https://dgalor.github.io/hist2style/ dgalor.github.io/hist2style/ .

Generative diffusion priors have recently achieved state-of-the-art performance in natural image super-resolution, demonstrating a powerful capability to synthesize photorealistic details. However, their direct application to remote sensing image super-resolution (RSISR) reveals significant shortcomings. Unlike natural images, remote sensing images exhibit a unique texture distribution where ground objects are globally stochastic yet locally clustered, leading to highly imbalanced textures. This imbalance severely hinders the model's spatial perception. To address this, we propose TexADiff, a novel framework that begins by estimating a Relative Texture Density Map (RTDM) to represent the texture distribution. TexADiff then leverages this RTDM in three synergistic ways: as an explicit spatial conditioning to guide the diffusion process, as a loss modulation term to prioritize texture-rich regions, and as a dynamic adapter for the sampling schedule. These modifications are designed to endow the model with explicit texture-aware capabilities. Experiments demonstrate that TexADiff achieves superior or competitive quantitative metrics. Furthermore, qualitative results show that our model generates faithful high-frequency details while effectively suppressing texture hallucinations. This improved reconstruction quality also results in significant gains in downstream task performance. The source code of our method can be found at https://github.com/ZezFuture/TexAdiff.

Diffusion models have found extensive use in solving inverse problems, by sampling from an approximate posterior distribution of data given the measurements. Recently, consistency models (CMs) have been proposed to directly predict the final output from any point on the diffusion ODE trajectory, enabling high-quality sampling in just a few neural function evaluations (NFEs). CMs have also been utilized for inverse problems, but existing CM-based solvers either require additional task-specific training or utilize data fidelity operations with slow convergence, limiting their applicability to large-scale problems and making them difficult to extend to nonlinear settings. In this work, we reinterpret CMs as proximal operators of a prior, enabling their integration into plug-and-play (PnP) frameworks. Specifically, we propose PnP-CM, an ADMM-based PnP solver that provides a unified framework for solving a wide range of inverse problems, and incorporates noise perturbations and momentum-based updates to improve performance in the low-NFE regime. We evaluate our approach on a diverse set of linear and nonlinear inverse problems. We also train and apply CMs to MRI data for the first time. Our results show that PnP-CM achieves high-quality reconstructions in as few as 4 NFEs, and produces meaningful results in 2 steps, highlighting its effectiveness in real-world inverse problems while outperforming existing CM-based approaches.

Vision-Language-Action (VLA) models provide a promising paradigm for robot learning by integrating visual perception with language-guided policy learning. However, most existing approaches rely on 2D visual inputs to perform actions in 3D physical environments, creating a significant gap between perception and action grounding. To bridge this gap, we propose a Spatial-Aware VLA Pretraining paradigm that enables models to acquire 3D spatial understanding before robot policy learning. Starting from pretrained vision-language models, we leverage large-scale human demonstration videos to extract 3D visual and 3D action annotations, forming a new source of supervision that aligns 2D visual observations with 3D spatial reasoning. We instantiate this paradigm with VIPA-VLA, a dual-encoder architecture that incorporates a 3D visual encoder to augment semantic visual representations with 3D-aware features, and aligns the two through visual-physical alignment pretraining. When adapted to downstream robot tasks, VIPA-VLA achieves significantly improved grounding between 2D vision and 3D action, resulting in more robust and generalizable robotic policies.

With the success of static black-hole imaging, the next frontier is the dynamic and 3D imaging of black holes. Recovering the dynamic 3D gas near a black hole would reveal previously-unseen parts of the universe and inform new physics models. However, only sparse radio measurements from a single viewpoint are possible, making the dynamic 3D reconstruction problem significantly ill-posed. Previously, BH-NeRF addressed the ill-posed problem by assuming Keplerian dynamics of the gas, but this assumption breaks down near the black hole, where the strong gravitational pull of the black hole and increased electromagnetic activity complicate fluid dynamics. To overcome the restrictive assumptions of BH-NeRF, we propose PI-DEF, a physics-informed approach that uses differentiable neural rendering to fit a 4D (time + 3D) emissivity field given EHT measurements. Our approach jointly reconstructs the 3D velocity field with the 4D emissivity field and enforces the velocity as a soft constraint on the dynamics of the emissivity. In experiments on simulated data, we find significantly improved reconstruction accuracy over both BH-NeRF and a physics-agnostic approach. We demonstrate how our method may be used to estimate other physics parameters of the black hole, such as its spin.

Lifelong Person Re-Identification aims to adapt to new domains while preserving old knowledge. Existing methods, whether distillation-based or rehearsal-based, attempt to consolidate diverse knowledge within a fixed model architecture. However, the limited adaptability of such architectures often leads to catastrophic forgetting by overwriting previously acquired knowledge. To overcome this limitation, we propose Versatile Incremental Adaptation (VIA), a novel dynamical expansion framework for LReID, which unleashes restricted knowledge during continuous learning by large pre-trained models. Specifically, Unseen-domain person Adapter (UnA) is embedded in VIA, which employs incremental modular learning to capture the domain's specific knowledge, thereby reducing cross-domain interference and releasing task-specific capacity that is previously limited by static parameter sharing. Meanwhile, considering the substantial amount of sharing knowledge across domains in LReID, we design the Domain-aware Dispatch module to enable inter-domain collaboration and knowledge reuse through adaptive cooperation among multiple shared adapters. Besides, a Holistic Domain Controller is designed to dynamically regulate the learning capacity for new domains based on knowledge similarity, thereby effectively unleashing the generalization potential of pre-trained models. Furthermore, a lightweight Similarity-Guided Auto-Selector is proposed to assign inputs to relevant adapters during inference automatically. Extensive experiments validate VIA's effectiveness, surpassing state-of-the-art methods in both seen and unseen domains.

Model merging integrates multiple task-specific models into a single consolidated one. Recent research has made progress in improving merging performance for in-distribution or multi-task scenarios, but domain generalization in model merging remains underexplored. We investigate how merging models fine-tuned on distinct domains affects generalization to unseen domains. Through an analysis of parameter competition in the task matrix using singular value decomposition, we show that merging models trained under different distribution shifts induces stronger conflicts between their subspaces compared to traditional multi-task settings. To mitigate this issue, we propose SCORE (Subspace COnflict-Resolving mErging), a method designed to alleviate such singular subspace conflicts. SCORE finds a shared orthogonal basis by computing the principal components of the concatenated leading singular vectors of all models. It then projects each task matrix into the shared basis, pruning off-diagonal components to remove conflicting singular directions. SCORE consistently outperforms, on average, existing model merging approaches in domain generalization settings across a variety of architectures and model scales, demonstrating its effectiveness and scalability.

Building generalist embodied agents requires a unified system that can interpret multimodal goals, model environment dynamics, and execute reliable actions across diverse real-world tasks. Multimodal large language models (MLLMs) offer strong semantic priors and cross-modal generalization, while world models (WMs) provide actionable latent dynamics for prediction and control. Their combination holds promise for open-ended embodied intelligence, yet introduces two key challenges: (1) establishing a tight coupling between the semantic intent from MLLMs and the dynamic state representations within the WM's latent space, and (2) achieving task-aware adaptability that supports multi-task learning and cross-environment generalization. To address these limitations, we propose ModularAgent, a task-aware dynamic joint framework that enables bidirectional coupling between MLLMs and WMs. ModularAgent establishes two complementary pathways: a forward path that injects MLLM representations into the WMs latent space for semantically guided imagination, and a backward path where WM-generated feedback refines the MLLMs semantic space via dense text-conditioned rewards. This bidirectional interaction is realized through three synergistic components: Task-Aware Dynamic Joint Learning, Task-Aware Behavior Learning, and MLLM-WM Joint Optimization, which together harmonize semantic reasoning and dynamic prediction. Extensive experiments across multi-task and cross-environment settings demonstrate superior stability and generalization over state-of-the-art baselines, marking a step toward open-ended embodied learning.

Robust 3D semantic occupancy is essential for legged and humanoid robots, yet most Semantic Scene Completion (SSC) systems are built for wheeled platforms with forward-facing sensors. We present OneOcc, a vision-only panoramic SSC framework tailored to severe body jitter and 360deg continuity. OneOcc integrates four complementary modules: (i) Dual-Projection fusion (DP-ER), which jointly exploits the raw annular panorama and its equirectangular unfolding to preserve true 360deg continuity while enabling grid-aligned feature extraction and seam-aware context; (ii) Bi-Grid Voxelization (BGV), which reasons in Cartesian and polar/cylindrical voxel spaces to reduce discretization bias and better align with panoramic geometry, yielding sharper free/occupied boundaries; (iii) a lightweight decoder with Hierarchical AMoE-3D fusion that dynamically routes multi-scale 3D features to specialized experts, improving long-range context and occlusion handling; and (iv) a plug-and-play Gait Displacement Compensation (GDC) module that learns feature-level motion correction from gait, stabilizing representations without extra sensors. We also release two panoramic occupancy benchmarks: QuadOcc (real quadruped, first-person 360deg) and Human360Occ (H3O) (CARLA human-ego 360deg with RGB/Depth/semantic-occupancy and standardized within-/cross-city splits). OneOcc sets new SOTA: on QuadOcc it exceeds strong vision baselines and even popular LiDAR methods, and on H3O it improves within-city by +3.83 mIoU and cross-city by +8.08. The modules are lightweight, enabling deployable full-surround semantic perception for legged and humanoid robots.

Modern LiDARs are rapidly transitioning from bulky, mechanically scanned systems to ultra-compact, low-cost, solid-state arrays. This miniaturization--while enabling scalability, affordability, and camera-like data structures--introduces a new and severe failure mode: internal-multipath glare. When light from a bright or retroreflective surface reflects and scatters within the LiDAR, light that should reach a single pixel spreads across the pixel array. The resulting artifacts create phantom objects, obscure real ones, and produce safety-critical "ghosts in the point clouds." This paper introduces a physically grounded sensing model and algorithmic techniques for addressing this effect. We show that internal glare can be represented as a linear, scene-independent operator--the Transient Glare Spread Function (TGSF)--acting on the transient measurements. Building on this model, we develop a training-free approach that operates on low-level LiDAR detections (or echoes) prior to point-cloud formation, leveraging knowledge of the glare spread function to reason about the likelihood of each detection arising from glare. The resulting approach is compatible with existing LiDAR signal-processing pipelines, and deployable on unmodified commercial sensors. Using experiments with real single-photon LiDAR hardware, we demonstrate substantial suppression of severe glare artifacts while preserving true scene structure.

Knowledge-based visual question answering (KB-VQA) demonstrates significant potential for handling knowledge-intensive tasks. However, conflicts arise between static parametric knowledge in vision language models (VLMs) and dynamically retrieved information due to the static model knowledge from pre-training. The outputs either ignore retrieved contexts or exhibit inconsistent integration with parametric knowledge, posing substantial challenges for KB-VQA. Current knowledge conflict mitigation methods primarily adapted from language-based approaches, focusing on context-level conflicts through engineered prompting strategies or context-aware decoding mechanisms. However, these methods neglect the critical role of visual information in conflicts and suffer from redundant retrieved contexts, which impair accurate conflict identification and effective mitigation. To address these limitations, we propose CC-VQA: a novel training-free, conflict- and correlation-aware method for KB-VQA. Our method comprises two core components: (1) Vision-Centric Contextual Conflict Reasoning, which performs visual-semantic conflict analysis across internal and external knowledge contexts; and (2) Correlation-Guided Encoding and Decoding, featuring positional encoding compression for low-correlation statements and adaptive decoding using correlation-weighted conflict scoring. Extensive evaluations on E-VQA, InfoSeek, and OK-VQA benchmarks demonstrate that CC-VQA achieves state-of-the-art performance, yielding absolute accuracy improvements of 3.3% to 6.4% compared to existing methods. Code will be released upon acceptance.

With the rapid advancement and widespread application of vision-language pre-training (VLP) models, their vulnerability to adversarial attacks has become a critical concern. In general, the adversarial examples can typically be designed to exhibit transferable power, attacking not only different models but also across diverse tasks. However, existing attacks on language-vision models mainly rely on static cross-modal interactions and focus solely on disrupting positive image-text pairs, resulting in limited cross-modal disruption and poor transferability. To address this issue, we propose a Semantic-Augmented Dynamic Contrastive Attack (SADCA) that enhances adversarial transferability through progressive and semantically guided perturbation. SADCA progressively disrupts cross-modal alignment through dynamic interactions between adversarial images and texts. This is accomplished by SADCA establishing a contrastive learning mechanism involving adversarial, positive and negative samples, to reinforce the semantic inconsistency of the obtained perturbations. Moreover, we empirically find that input transformations commonly used in traditional transfer-based attacks also benefit VLPs, which motivates a semantic augmentation module that increases the diversity and generalization of adversarial examples. Extensive experiments on multiple datasets and models demonstrate that SADCA significantly improves adversarial transferability and consistently surpasses state-of-the-art methods. The code is released at https://github.com/LiYuanBoJNU/SADCA.

Few-shot learning has attracted extensive attention, with metric-based approaches such as Prototypical Networks establishing strong baselines. These methods construct class prototypes from support samples and classify query samples via distance metrics, but their performance is highly sensitive to label noise. To tackle this challenge, we propose a novel Graph Attention Prototypical Network (GAPNet) for robust few-shot classification. GAPNet first extracts local and global features via a classic CNN backbone and a group attention broad learning module, respectively. To mitigate the impact of label noise, the intra-class and inter-class relationships between support and query samples are explicitly modeled via a pseudo-label guided graph constructor, and then processed by an edge-aware graph attention module to capture topological correlations. Furthermore, an adaptive noise-robust prototype generator is introduced to dynamically suppress the contributions of noisy samples, substantially improving the reliability of class prototypes. Extensive experiments demonstrate the effectiveness and robustness of GAPNet to label noise. Compared to state-of-the-art approaches, GAPNet improves accuracy in the 5-way 5-shot setting by 3% ~ 8% on three general image benchmarks and one fine-grained classification dataset.

Large Vision-Language Models (LVLMs) often omit or misrepresent critical visual content in generated image captions. Minimizing such information loss will force LVLMs to focus on image details to generate precise descriptions. However, measuring information loss during modality conversion is inherently challenging due to the modal gap between visual content and text output. In this paper, we argue that the quality of an image caption is positively correlated with the similarity between images retrieved via text search using that caption. Based on this insight, we further propose Cross-modal Identity Mapping (CIM), a reinforcement learning framework that enhances image captioning without requiring additional annotations. Specifically, the method quantitatively evaluates the information loss from two perspectives: Gallery Representation Consistency and Query-gallery Image Relevance. Supervised under these metrics, LVLM minimizes information loss and aims to achieve identity mapping from images to captions. The experimental results demonstrate the superior performance of our method in image captioning, even when compared with Supervised Fine-Tuning. Particularly, on the COCO-LN500 benchmark, CIM achieves a 20% improvement in relation reasoning on Qwen2.5-VL-7B. The codes are released on GitHub.

Achieving strong policy generalization to unseen environments remains a core challenge in visual reinforcement learning, and segmenting task-relevant regions to mitigate the influence of irrelevant visual cues has emerged as a promising direction. However, existing methods rely solely on the current observation, lack temporal information, and fail to exploit preceding observations, leaving learned policies susceptible to task-irrelevant background variations and ultimately degrading policy performance. In this paper, we propose temporal segmentation for task-relevant mask in visual reinforcement learning, named TSTM, which extracts task-relevant regions from sequential observations by exploiting temporal information, thereby producing more reliable masks and improving policy generalization. TSTM introduces a temporal segmentation network with an encoder-temporal-decoder architecture, where a convolutional LSTM module captures temporal dependencies across observations. To reduce inference overhead, we further develop a lightweight student network as an efficient substitute for the teacher network. The resulting task-relevant masks are encoded by a CNN-based encoder, and invariant representation learning is employed to improve robustness by enforcing consistency between representations of the original and augmented observation sequences. With these task-relevant representations, we train an actor-critic agent to learn a policy with strong generalization capability. Experimental results demonstrate that TSTM achieves superior generalization performance over existing state-of-the-art methods on most visual RL tasks.

Vision-Language Models (VLMs) have enhanced traditional LLMs with visual capabilities through the integration of vision encoders. While recent works have explored various combinations of vision encoders and LLMs, there still lacks a principled understanding of what makes a vision encoder suitable for VLM alignment. In this paper, we systematically investigate this question via comprehensive experiments on a curated collection of 19 pre-trained vision encoders from diverse sources. We first demonstrate that common practices, such as choosing encoders with the largest size or highest zero-shot accuracy, consistently fail to identify optimal models. In fact, these metrics show only weak to moderate correlation with VLM performance. This intriguing finding begs a fundamental question: What factors of vision-encoders matter in VLM? Through comprehensive analysis, we identify that the structural similarity across modalities plays a crucial but previously overlooked role in vision-encoder selection, which we measure using the Gromov-Wasserstein distance as a proxy. From a theoretical perspective, we show that the learnability of cross-modality mapping can be provably associated with the Gromov-Wasserstein distance. Empirical verification on 60+ full VLM training runs shows that our proposed inference-only metric performs significantly better than alternative model selection strategies and exhibits a much stronger correlation with final VLM performance, thereby enabling efficient and effective prediction of VLM performance before full training.