CVPR 2026 论文列表

Text-to-image (T2I) models are increasingly popular, producing a large share of AI-generated images online. To compare model quality, voting-based leaderboards have become the standard, relying on anonymized model outputs for fairness. In this work, we show that such anonymity can be easily broken. We find that generations from each T2I model form distinctive clusters in the image embedding space, enabling accurate deanonymization without prompt control or training data. Using 22 models and 280 prompts (150K images), our centroid-based method achieves high accuracy and reveals systematic model-specific signatures. We further introduce a prompt-level distinguishability metric and conduct large-scale analyses showing how certain prompts can lead to near-perfect distinguishability. Our findings expose fundamental security flaws in T2I leaderboards and motivate stronger anonymization defenses.

Due to the lack of effective cross-modal modeling, existing open-source audio-video generation methods often exhibit compromised lip synchronization and insufficient semantic consistency. To mitigate these drawbacks, we propose UniAVGen, a unified framework for human-centric joint audio and video generation. UniAVGen is anchored in a dual-branch joint synthesis architecture, incorporating two parallel Diffusion Transformers (DiTs) to build a cohesive cross-modal latent space. At its heart lies an Asymmetric Cross-Modal Interaction mechanism, which enables bidirectional, temporally aligned cross-attention, thus ensuring precise spatiotemporal synchronization and semantic consistency. Furthermore, this cross-modal interaction is augmented by a Face-Aware Modulation (FAM) module, which dynamically prioritizes salient regions in the interaction process. To enhance generative fidelity during inference, we additionally introduce Modality-Aware Classifier-Free Guidance (MA-CFG), a novel strategy that explicitly amplifies cross-modal correlation signals. Notably, UniAVGen's robust joint synthesis design enables the seamless unification of pivotal audio-visual tasks within a single model. Furthermore, we demonstrate that joint multi-task training can further boost the performance of joint generation. Comprehensive experiments validate that, with far fewer training samples (1.3M vs. 30.1M), UniAVGen delivers overall advantages in audio-video synchronization, timbre consistency, and emotion consistency.

Recent image editing models boast next-level intelligent capabilities, facilitating cognition- and creativity-informed image editing. Yet, existing benchmarks provide too narrow a scope for evaluation, failing to holistically assess these advanced abilities. To address this, we introduce WiseEdit, a knowledge-intensive benchmark for comprehensive evaluation of cognition- and creativity-informed image editing, featuring deep task depth and broad knowledge breadth. Drawing an analogy to human cognitive creation, WiseEdit decomposes image editing into three cascaded steps--Awareness, Interpretation, and Imagination--each corresponding to a task that poses a challenge for models to complete at the specific step. It also encompasses complex tasks, where none of the three steps can be finished easily. Furthermore, WiseEdit incorporates three fundamental types of knowledge: Declarative, Procedural, and Metacognitive knowledge. Ultimately, WiseEdit comprises 1,220 test cases, objectively revealing the limitations of SoTA image editing models in knowledge-based cognitive reasoning and creative composition capabilities.

Text-to-image (T2I) models such as Stable Diffusion and DALLE remain susceptible to generating harmful or Not-Safe-For-Work (NSFW) content under jailbreak attacks despite deployed safety filters. Existing jailbreak attacks either rely on proxy-loss optimization instead of the true end-to-end objective, or depend on large-scale and costly RL-trained generators. Motivated by these limitations, we propose JANUS , a lightweight framework that formulates jailbreak as optimizing a structured prompt distribution under a black-box, end-to-end reward from the T2I system and its safety filters. JANUS replaces a high-capacity generator with a low-dimensional mixing policy over two semantically anchored prompt distributions, enabling efficient exploration while preserving the target semantics. On modern T2I models, we outperform state-of-the-art jailbreak methods, improving ASR-8 from 25.30% to 43.15% on Stable Diffusion 3.5 Large Turbo with consistently higher CLIP and NSFW scores. JANUS succeeds across both open-source and commercial models. These findings expose structural weaknesses in current T2I safety pipelines and motivate stronger, distribution-aware defenses. Warning: This paper contains model outputs that may be offensive.

Reliable navigation and decision-making of autonomous vehicles require both accurate localization and object detection. Traditionally, these two tasks are handled separately, leading to redundant computation and limited cross-task knowledge transfer. This paper proposes TACO, the first Task-Aware COntrastive learning framework, which performs joint LiDAR localization and 3D object detection within a single, unified network. TACO leverages contrastive learning to explicitly decouple and align static geographic features for localization and object-centric features for detection. This bidirectional mutual supervision not only enhances localization robustness in dynamic environments by filtering dynamic noise but also boosts detection accuracy via effective spatial context. Additionally, we propose OxfoLD, the first dataset that provides multi-traversal LiDAR localization ground truth with rich 3D object annotations, thereby supporting task validation across various times and weather conditions. Experimental results demonstrate that TACO achieves state-of-the-art localization accuracy while maintaining competitive detection performance. The code and dataset will be released.

Recent advancements in instruction-based image editing and subject-driven generation have garnered significant attention, yet both tasks still face limitations in meeting practical user needs. Instruction-based editing relies solely on language instructions, which often fail to capture specific editing details, making reference images necessary. Meanwhile, subject-driven generation is limited to combining concrete objects or people, overlooking broader, abstract concepts. To address these challenges, we propose two novel tasks: multimodal instruction-based editing and generation. These tasks support both text and image instructions and extend the scope to include both concrete and abstract concepts, greatly enhancing their practical applications. We introduce DreamOmni2, tackling two primary challenges: data creation and model framework design. Our data synthesis pipeline consists of three steps: (1) using a feature mixing method to create extraction data for both abstract and concrete concepts, (2) generating multimodal instruction-based editing training data using the editing and extraction models, and (3) further applying the extraction model to create training data for multimodal instruction-based editing. For the framework, to handle multi-image input, we propose an index encoding and position encoding shift scheme, which helps the model distinguish images and avoid pixel confusion. Additionally, we introduce joint training with the VLM and our generation/editing model to better process complex instructions. In addition, we have proposed comprehensive benchmarks for these two new tasks to drive their development. Experiments show that DreamOmni2 achieved impressive results.

Color transfer aims to match the color distribution of a content image (source) to that of a style image (target) while preserving structure and perceptual realism. Yet modulation-based flow models such as ModFlows often produce trajectory misalignment and artifacts because they rely on strictly linear transport paths. We propose NCT, a nonlinear color transfer framework that replaces linear paths with Bezier trajectories, enabling smooth, nonlinear, and perceptually coherent color transfer. This parameterization lets the transport bend toward plausible intermediate color regimes, improving content-style alignment and reducing chromatic distortion. We further incorporate a Mixture of Experts (MoE) module in the encoder to select trajectory experts for different chromatic regimes, improving generalization to heterogeneous data with complex illumination and materials. Experiments show that NCT reduces artifacts and achieves more stable color transfer than prior flow-based methods, especially on 3D-rendered or highly textured images. The code is provided in supplementary materials.

Current foundation models for 3D shapes excel at global tasks (retrieval, classification) but transfer poorly to local part-level reasoning. Recent approaches leverage vision and language foundation models to directly solve dense tasks through multi-view renderings and text queries. While promising, these pipelines require expensive inference over multiple renderings, depend heavily on large language-model (LLM) prompt engineering for captions, and fail to exploit the inherent 3D geometry of shapes. We address this gap by introducing an encoder-only 3D model that produces language-aligned patch-level features directly from point clouds. Our pre-training approach builds on existing data engines that generate part-annotated 3D shapes by pairing multi-view SAM regions with VLM captioning. Using this data, we train a point cloud transformer encoder in two stages: (1) distillation of dense 2D features from visual encoders such as DINOv2 into 3D patches, and (2) alignment of these patch embeddings with part-level text embeddings through a multi-positive contrastive objective. Our 3D encoder achieves zero-shot 3D part segmentation with fast single-pass inference without any test-time multi-view rendering, while significantly outperforming previous rendering-based and feed-forward approaches across several 3D part segmentation benchmarks.

Most of the models used to generate embeddings for retrieval are not trained for the purpose which leads them to focus on coarse semantic alignment rather than particular object attributes or arrangements. This limits their performance, particularly on challenging problems such as cross-modal fine-grained retrieval. Furthermore, their training objectives lack the discriminative ability required to distinguish between descriptions that are semantically similar but factually different. To address these challenges, we propose POGA (Paraphrased and Oppositional Graph Alignment), a novel framework for fine-grained cross-modal alignment. POGA comprises two core innovations: (1) Multi-source Graph Augmentation (MSGA), which not only generates paraphrased positives and oppositional negatives, but also parses the image and all text variants into structured graphs to provide difference-rich supervisory signals; (2) Hybrid Multi-granularity Alignment (HMA), which defines a composite training objective that jointly optimizes the model at four distinct granularities: including robust dual global alignment, and precise matching at three fine-grained levels: node, relation, and focal disproving. Experiments on benchmarks such as DCI and DOCCI demonstrate that POGA performs favorably against several state-of-the-art methods in long-text understanding and complex relation discrimination.

Forecasting dynamic scenes remains a fundamental challenge in computer vision, as limited observations make it difficult to capture coherent object-level motion and long-term temporal evolution.We present Motion Group-aware Gaussian Forecasting (MoGaF), a framework for long-term scene extrapolation built upon the 4D Gaussian Splatting representation.MoGaF introduces motion-aware Gaussian grouping and group-wise optimization to enforce physically consistent motion across both rigid and non-rigid regions, yielding spatially coherent dynamic representations.Leveraging this structured space-time representation, a lightweight forecasting module predicts future motion, enabling realistic and temporally stable scene evolution.Experiments on synthetic and real-world datasets demonstrate that MoGaF consistently outperforms existing baselines in rendering quality, motion plausibility, and long-term forecasting stability.

Retrieving rare and safety-critical driving scenarios from large-scale datasets is essential for building robust autonomous driving (AD) systems. As dataset sizes continue to grow, the key challenge shifts from collecting more data to efficiently identifying the most relevant samples. We introduce SearchAD, a large-scale rare image retrieval dataset for AD containing over 423k frames drawn from 11 established datasets. SearchAD provides high-quality manual annotations of more than 513k bounding boxes covering 90 rare categories. It specifically targets the "needle-in-a-haystack" problem of locating extremely rare classes, with some appearing fewer than 50 times across the entire dataset. Unlike existing benchmarks, which focused on instance-level retrieval, SearchAD emphasizes semantic image retrieval with a well-defined data split, enabling text-to-image and image-to-image retrieval, few-shot learning, and fine-tuning of multi-modal retrieval models. Comprehensive evaluations show that text-based methods outperform image-based ones due to stronger inherent semantic grounding. While models directly aligning spatial visual features with language achieve the best zero-shot results, and our fine-tuning baseline significantly improves performance, absolute retrieval capabilities remain unsatisfactory. With a held-out test set on a public benchmark server, SearchAD establishes the first large-scale dataset for retrieval-driven data curation and long-tail perception research in AD: https://iis-esslingen.github.io/searchad/

Directly editing ultra-high-resolution (UHR) images is valuable but underexplored, primarily due to the lack of high-quality data and the challenge in modeling high-frequency texture details. We introduce VINS-120K, the first large-scale dataset for instruction-based UHR image editing, comprising 120K carefully curated triplets of instruction, input image, and edited image. Each image exceeds 4K resolution (\geq4096x4096) and is filtered through a rigorous multi-stage pipeline to ensure visual quality, instruction alignment, and aesthetic fidelity. Built on VINS-120K, we further develop a high-frequency-aware post-adaptation strategy to extend pretrained non-high-resolution models to the UHR regime. We also present VINS-4KEval, a benchmark covering diverse editing types, to facilitate consistent evaluation in UHR settings. Experiments confirm that our work improves fine-grained detail synthesis and texture realism in UHR image editing.

Despite the remarkable progress in open-vocabulary object detection (OVD), a significant gap remains between the training and testing phases. During training, the RPN and RoI heads often misclassify unlabeled novel-category objects as background, causing some proposals to be prematurely filtered out by the RPN while others are further misclassified by the RoI head. During testing, these proposals again receive low scores and are removed in post-processing, leading to a significant drop in recall and ultimately weakening novel-category detection performance.To address these issues, we propose a novel training framework--NoOVD--which innovatively integrates a self-distillation mechanism grounded in the knowledge of frozen vision-language models (VLMs).Specifically, we design K-FPN, which leverages the pretrained knowledge of VLMs to guide the model in discovering novel-category objects and facilitates knowledge distillation--without requiring additional data--thus preventing forced alignment of novel objects with background.Additionally, we introduce R-RPN, which adjusts the confidence scores of proposals during inference to improve the recall of novel-category objects.Cross-dataset evaluations on OV-LVIS, OV-COCO, and Objects365 demonstrate that our approach consistently achieves superior performance across multiple metrics.

Dexterous manipulation is essential for real-world robot autonomy, mirroring the central role of human hand coordination in daily activity. Humans rely on rich multimodal perception--vision, sound, and language-guided intent--to perform dexterous actions, motivating vision-based, language-conditioned manipulation systems for robots. However, training reliable vision-language-action (VLA) models for dexterous manipulation requires large-scale demonstrations across many robotic hands. In addition, as new dexterous embodiments appear rapidly, collecting data for each becomes costly and impractical, creating a need for scalable cross-embodiment learning. We introduce \ourmethod, a vision-language-action framework integrated with a unified latent action space shared across diverse dexterous hands. This embodiment-invariant latent space is directly pluggable into standard VLA architectures, enabling seamless cross-embodiment training and efficient reuse of both existing and newly collected data. Experimental results demonstrate that \ourmethod consistently outperforms baseline VLA models operating in raw joint spaces, establishing it as an effective solution for scalable cross-embodiment dexterous manipulation.

Post-training of large-scale Vision-Language Models (VLMs) reveals a pronounced generalization gap: models fine-tuned with Reinforcement Learning (RL) consistently achieve superior out-of-distribution (OOD) performance compared to those trained with Supervised Fine-Tuning (SFT). This paper posits a data-centric explanation for this phenomenon, contending that RL's generalization advantage arises from an implicit data filtering mechanism that inherently prioritizes medium-difficulty training samples. To test this hypothesis, we systematically evaluate the OOD generalization of SFT models across training datasets of varying difficulty levels. Our results confirm that data difficulty is a critical factor, revealing that training on hard samples significantly degrades OOD performance. Motivated by this finding, we introduce Difficulty-Curated SFT (DC-SFT), a straightforward method that explicitly filters the training set based on sample difficulty. Experiments show that DC-SFT not only substantially enhances OOD generalization over standard SFT, but also surpasses the performance of RL-based training, all while providing greater stability and computational efficiency. This work offers a data-centric account of the OOD generalization gap in VLMs and establishes a more efficient pathway to achieving robust generalization. Code is available at https://github.com/byyx666/DC-SFT.

Gait recognition, as a reliable biometric technology, has seen rapid development in recent years while it faces significant challenges caused by diverse clothing styles in the real world. This paper introduces BarbieGait, a synthetic gait dataset where real-world subjects are uniquely mapped into a virtual engine to simulate extensive clothing changes while preserving their gait identity information. As a pioneering work, BarbieGait provides a controllable gait data generation method, enabling the production of large datasets to validate cross-clothing issues that are difficult to verify with real-world data. However, the diversity of clothing increases intra-class variance and makes one of the biggest challenges to learning cloth-invariant features under varying clothing conditions. Therefore, we propose GaitCLIF (Gait-oriented CLoth-Invariant Feature) as a robust baseline model for cross-clothing gait recognition. Through extensive experiments, we validate that our method significantly improves cross-clothing performance on BarbieGait and the existing popular gait benchmarks. We believe that BarbieGait, with its extensive cross-clothing gait data, will further advance the capabilities of gait recognition in cross-clothing scenarios and promote progress in related research. The source code and dataset are available at https://github.com/BarbieGait/BarbieGait.

Mixture-of-Experts (MoE) Multimodal large language models (MLLMs) excel at vision-language tasks, but they suffer from high computational inefficiency. To reduce inference overhead, expert skipping methods have been proposed to deactivate redundant experts based on the current input tokens. However, we find that applying these methods--originally designed for unimodal large language models (LLMs)--to MLLMs results in considerable performance degradation. This is primarily because such methods fail to account for the heterogeneous contributions of experts across MoE layers and modality-specific behaviors of tokens within these layers. Motivated by these findings, we propose MoDES, the first training-free framework that adaptively skips experts to enable efficient and accurate MoE MLLM inference. It incorporates a globally-modulated local gating (GMLG) mechanism that integrates global layer-wise importance into local routing probabilities to accurately estimate per-token expert importance. A dual-modality thresholding (DMT) method is then applied, which processes tokens from each modality separately, to derive the skipping schedule. To set the optimal thresholds, we introduce a frontier search algorithm that exploits monotonicity properties, cutting convergence time from several days to a few hours. Extensive experiments for 3 model series across 13 benchmarks demonstrate that MoDES far outperforms previous approaches. For instance, when skipping 88% experts for Qwen3-VL-MoE-30B-A3B-Instruct, the performance boost is up to **10.67%** (97.33% vs. 86.66%). Furthermore, MoDES significantly enhances inference speed, improving the prefilling time by **2.16x** and the decoding time by **1.26x**.

With the rapid growth of stereoscopic 3D devices, real-time stereoscopic conversion has become increasingly essential. However, most existing approach rely on depth estimation, forward warping, and heavy inpainting network, resulting in high computational cost and artifacts near occlusion boundaries. Diffusion-based models have also been explored, but they suffer from iterative sampling and geometric inconsistency, making them unsuitable for real-time deployment. To address these issues, we propose Bi-Warp, a simple yet effective approach that synthesizes the right view without inpainting network by leveraging warping operations. Our approach estimates backward flow, approximates the corresponding forward flow, and generates two candidate right views via bidirectional warping. A learnable mask adaptively fuses the candidates, preserving left-right geometric consistency. Building on Bi-Warp, we introduce XPaintNet, a lightweight network that achieves comparable visual quality to state-of-the-art methods while maintaining real-time performance over 100 FPS at 2K resolution.

We introduce BuildAnyPoint, a novel generative framework for structured 3D building reconstruction from point clouds with diverse distributions, such as those captured by airborne LiDAR and Structure-from-Motion.To recover artist-created building abstraction in this highly underconstrained setting, we capitalize on the role of explicit 3D generative priors in autoregressive mesh generation.Specifically, we design a Loosely Cascaded Diffusion Transformer (Loca-DiT) that initially recovers the underlying distribution from noisy or sparse points, followed by autoregressively encapsulating them into compact meshes.We first formulate distribution recovery as a conditional generation task by training latent diffusion models conditioned on input point clouds, and then tailor a decoder-only transformer for conditional autoregressive mesh generation based on the recovered point clouds.Our method delivers substantial qualitative and quantitative improvements over prior building abstraction methods. Furthermore, the effectiveness of our approach is evidenced by the strong performance of its recovered point clouds on building point cloud completion benchmarks, which exhibit improved surface accuracy and distribution uniformity.