CVPR 2026 Papers

This paper introduces UniMERNet, a high-accuracy, computation-efficient algorithm for Mathematical Expression Recognition (MER) across diverse real-world scenarios. To facilitate UniMERNet's training, we constructed UniMER-1M, a million-scale dataset whose unprecedented diversity endows the model with robust generalization ability. Through in-depth analysis, we discover a distinctive raster-scan pattern (left-to-right, top-to-bottom) in the attention distribution of Transformer models for MER tasks, which closely aligns with human reading habits. Based on this key finding, we design an innovative **Raster-Scan Attention** mechanism that employs a "horizontal-first, vertical-second" sequential attention computation strategy. This approach not only successfully reduces computational complexity from \mathcal O (NH^2W^2D) to \mathcal O (NHWD(H + W)) , but also enables the model to capture long-range dependencies more efficiently, achieving recognition performance comparable to global attention. Leveraging both UniMER-1M and our innovative attention mechanism, UniMERNet achieves state-of-the-art performance across four real-world scenarios while significantly reducing computational resources compared to global attention: over **1.2x** memory savings during training, approximately **10x** memory reduction during inference, and **5x** speed with slightly improved accuracy. All resources will be publicly released to advance MER research further.

As recent AI models have successfully generated high-resolution photorealistic images, it has also been socially important to detect whether an image is generated by AI. Since training data for the detection task is often not available due to the diversity of generative models, training-free detection approaches have been practically considered. A common approach is to utilize the image-level reconstruction error from the latent diffusion model (LDM). However, we find this score suffers from instance-specific biases, particularly in images with simple backgrounds. To this end, we propose a novel image-level debiasing score function that cancels out background contribution by normalizing the reconstruction error on the augmented images with similar background information. To be specific, we show that rotation and low-pass filtering are effective augmentation strategies. To promote generalization to broader generative models, we newly explore latent-level reconstruction error as an additional training-free signal. However, we observe that the latent-level score also suffers to latent-specific bias. To mitigate this, we introduce a rotation-based latent-level debiasing score based on the normalization of the rotated latent. We unify the aforementioned scores into a single unified debiasing score, RDD, which achieves state-of-the-art training-free detection performance across diverse generative models. Furthermore, our framework can be robust to corruption of the examined images.

3D Gaussian Splatting (3DGS) has revolutionized radiance field reconstruction by achieving high-quality novel view synthesis with fast rendering speed, introducing 3D Gaussian primitives to represent the scene. However, 3DGS encounters blurring and floaters when applied to complex scenes, caused by the reconstruction of redundant and ambiguous geometric structures. We attribute this issue to the unstable optimization of the Gaussians. To address this limitation, we present a plug-and-play PDE-based optimization method that overcomes the optimization constraints of 3DGS-based approaches in various tasks, such as novel view synthesis and surface reconstruction. Firstly, we theoretically derive that the 3DGS optimization procedure can be modeled as a PDE, and introduce a viscous term to ensure stable optimization. Secondly, we use the Material Point Method (MPM) to obtain a stable numerical solution of the PDE, which enhances both global and local constraints. Additionally, an effective Gaussian densification strategy and particle constraints are introduced to ensure fine-grained details. Extensive qualitative and quantitative experiments confirm that our method achieves state-of-the-art rendering and reconstruction quality.

Vision-language models (VLMs) have achieved remarkable success across diverse tasks. However, concerns about their trustworthiness persist, particularly regarding tendencies to lean more on textual cues than visual evidence and the risk of producing ungrounded or fabricated responses. To address these issues, we propose Saliency-R1, a framework for improving the interpretability and faithfulness of VLMs reasoning. Specifically, we introduce a novel saliency map technique that efficiently highlights critical image regions contributing to generated tokens without additional computational overhead. This can further be extended to trace how visual information flows through the reasoning process to the final answers, revealing the alignment between the thinking process and the visual context. We use the overlap between the saliency maps and human-annotated bounding boxes as the reward function, and apply Group Relative Policy Optimization (GRPO) to align the salient parts and critical regions, encouraging models to focus on relevant areas when conduct reasoning. Experiments show Saliency-R1 improves reasoning faithfulness, interpretability, and overall task performance.

Diffusion-based video editing has emerged as an important paradigm for high-quality and flexible content generation. However, despite their generality and strong modeling capacity, Diffusion Transformers (DiT) remain computationally expensive due to the iterative denoising process, posing challenges for practical deployment. Existing video diffusion acceleration methods primarily exploit denoising timestep-level feature reuse, which mitigates the redundancy in denoising process, but overlooks the architectural redundancy within the DiT that many attention operations over spatio-temporal tokens are redundantly executed, offering little to no incremental contribution to the model's output. This work introduces HetCache, a training-free diffusion acceleration framework designed to exploit the inherent heterogeneity in diffusion-based masked video-to-video (MV2V) generation and editing. Instead of uniformly reuse or randomly sampling tokens, HetCache assesses the contextual relevance and interaction strength among various types of tokens in designated computing steps. Guided by spatial priors, it divides the spatial-temporal tokens in DiT model into context and generative tokens, and selectively caches the context tokens that exhibit the strongest correlation and most representative semantics with generative ones. This strategy reduces redundant attention operations while maintaining editing consistency and fidelity. Experiments show that HetCache achieves a noticeable acceleration, including a 2.67 xlatency speedup and FLOPs reduction over commonly used foundation models, with negligible degradation in editing quality.

Cross-modal biomedical signals such as pathology and genomics can provide richer and more robust semantic guidance for medical image representation learning. However, the availability of such guidance remains limited, as privacy constraints and acquisition costs severely restrict access to medical images paired with other biomedical data. A further challenge lies in modality discrepancy, which introduces intra-modal statistical bias and cross-modal noise, thereby degrading the quality of medical image representations. To address these challenges, we propose KAMP, a large language model (LLM)-driven multimodal pretraining framework for medical image representation learning. KAMP leverages textual priors as the semantic anchor to enhance medical image representations and align them with multimodal biomedical representations, enabling the learning of rich and robust features even when paired data are scarce. KAMP operates in three stages. First, the LLM generates personalized diagnostic knowledge from patient clinical text and imaging metadata. This knowledge is injected as a prior to enrich medical image representations and serves as a semantic anchor to reduce the representation gap between medical images and other biomedical modalities. Second, the LLM is optimized using Group Relative Policy Optimization (GRPO), with the cross-modal aligner pretrained in the first stage serving as the reward model. Third, the refined knowledge is used to retrain the cross-modal aligner, yielding more robust medical image representations while mitigating bias and noise introduced by other modalities. Comprehensive evaluations on brain, bladder, and liver cancer datasets demonstrate that KAMP outperforms existing methods in most downstream few-shot classification tasks.

While Video Large Language Models (Video-LLMs) have shown significant potential in multimodal understanding and reasoning tasks, how to efficiently select the most informative frames from videos remains a critical challenge. Existing methods attempt to optimize frame sampling by reducing inter-frame redundancy or employing unsupervised event localization. However, these approaches often fall short in handling complex instruction-following tasks and scenarios that demand precise temporal modeling, resulting in limited performance in both semantic alignment and temporal reasoning. To address the above challenges, we introduce Instructed Temporal Grounding for Videos (VideoITG), a framework aiming to adaptively customize frame sampling strategies based on user instructions. Specifically, we design the VidThinker pipeline, which automates annotation by generating instruction-conditioned captions, retrieving relevant video segments, and selecting key frames to enable efficient supervision. Using VidThinker, we build the VideoITG-40K dataset with 40K videos and 500K temporal grounding annotations. Our plug-and-play VideoITG model leverages Video-LLMs' visual-language alignment and reasoning for discriminative frame selection. VideoITG consistently boosts the performance on multiple multimodal video understanding benchmarks, demonstrating its effectiveness and potential.

Current 3D human animation methods fail at photorealism: kinematics-based approaches lack non-rigid dynamics like clothing, while methods reconstructing from generated videos suffer from low-quality artifacts and identity loss. To overcome these limitations, we present Ani3DHuman, a framework that marries kinematics-based animation with video diffusion priors. We first introduce a layered motion representation that disentangles rigid motion and residual non-rigid motion. Then, we use a pretrained video diffusion model to restore a coarse rendering from the mesh-rigged animation, which provides supervision for the motion field. However, this restoration task, based on diffusion sampling, is highly challenging, as the initial renderings are out-of-distribution, causing standard deterministic ODE samplers to fail. Therefore, our core technical contribution is self-guided stochastic sampling, which effectively solves the out-of-distribution problem by combining stochastic sampling (for photorealistic quality) with self-guidance (for identity fidelity). These restored videos provide high-quality supervision, enabling the optimization of a realistic 4D motion field. Ani3DHuman achieves state-of-the-art results, and our ablations validate that both components of our sampler are essential for high-fidelity restoration.

Reward models (RMs) are essential for training large language models (LLMs), but remain underexplored for omni models that handle interleaved image and text sequences. We introduce Multimodal RewardBench 2 (MMRB2), the first comprehensive benchmark for reward models on multimodal understanding and (interleaved) generation. MMRB2 spans four tasks: text-to-image, image editing, interleaved generation, and multimodal reasoning ("thinking-with-images"), providing 1,000 expert-annotated preference pairs per task from 23 models and agents across 21 source tasks. MMRB2 is designed with: (1) practical but challenging prompts; (2) responses from state-of-the-art models, and agents; and (3) preference pairs with strong human-expert consensus, curated via an ensemble filtering strategy. Using MMRB2, we study existing judges for each subtask, including multimodal LLM-as-a-judge and models trained with human preferences. Top judges like GPT-5 and Gemini-2.5-Pro reach 66-75% accuracy, compared to >90% for humans, and outperform the commonly used GPT-4o (59% accuracy). The best performing open-source model Qwen3-VL-32B achieves similar accuracies as Gemini-2.5-Flash (64%). We also show that MMRB2 performance strongly correlates with downstream task success, and conduct an in-depth analysis that shows key areas to improve the reward models going forward.

High-precision dichotomous image segmentation (DIS) is a task of extracting fine-grained objects from high-resolution images.Existing methods trade efficiency for accuracy: non-diffusion methods are fast but suffer from weak semantics and unstable spatial priors, causing false detections; diffusion-based methods offer high accuracy via strong generative priors but are computationally expensive.In depth maps, a complete object appears as a low variance region with a smooth interior and sharp boundaries, whereas the background exhibits a chaotic, high variance pattern due to disconnected surfaces at varying depths. We refer to this as the depth integrity-prior.Inspired by this, and noting that DIS currently lacks depth maps, we leverage pseudo-depth information from monocular depth estimation models to obtain essential semantic understanding, thereby rapidly revealing spatial differences across target objects and the background.To exploit this prior, we propose the Prior-guided Depth Fusion Network (PDFNet), which fuses RGB and pseudo-depth features for depth-aware structure perception. We further introduce a novel depth integrity-prior loss to enforce depth consistency in segmentation and a fine-grained enhancement module with adaptive patch selection to sharpen boundaries.Notably, PDFNet with DAM-v2 achieves SOTA (F^ max _b 0.915 on DIS-VD and 0.915 on DIS-TE) using less than half the params of diffusion-based methods.

In real-world robotic manipulation, states typically admit a neighborhood of near-equivalent actions. That is, for each state, there exist a feasible action neighborhood (FAN) rather than a single correct action, within which motions yield indistinguishable progress. However, prevalent VLA training methodologies are directly inherited from linguistic settings and do not exploit the FAN property, thus leading to poor generalization and low sample efficiency. To address this limitation, we introduce a FAN-guided regularizer that shapes the model's output distribution to align with the geometry of FAN. Concretely, we introduce a Gaussian prior that promotes locally smooth and unimodal predictions around the preferred direction and magnitude. In extensive experiments across both reinforced finetuning (RFT) and supervised finetuning (SFT), our method achieves significant improvement in sample efficiency, and success rate in both in-distribution and out-of-distribution (OOD) scenarios. By aligning with the intrinsic action tolerance of physical manipulation, FAN-guided regularization provides a principled and practical method for sample-efficient, and generalizable VLA adaptation. The source code is publicly available at: https://github.com/Haochen-Niu/FAN-VLA.

Few-shot fine-grained image classification (FSFG) aims to recognize novel fine-grained categories from only a few labeled samples. Existing FSFG methods primarily focus on fine-grained feature extraction and modeling query-support interactions within training episodes containing a small number of classes. Relying on the episodic training strategy, these methods typically assume that the capabilities learned on training samples can directly transfer to evaluation episodes with a few novel classes (few-way). However, in more practical and challenging scenarios involving many novel classes (many-way), existing approaches lack a reliable and global characterization of the feature space, making it difficult for episodic adaptation alone to generalize effectively. In this paper, we pioneer a theoretical analysis of novel class behavior in FSFG and derive a class discriminative index bound. Guided by this analysis, we propose a novel SCEG method that incorporates Self and Collaborative feature extraction as well as Episodic and Global feature space optimization. Extensive experiments demonstrate that our method consistently and significantly outperforms existing methods under both conventional few-way and the new many-way settings. Codes are available at: https://github.com/Legenddddd/SCEG.

Multiple Instance Learning (MIL) has emerged as the dominant paradigm for the analysis of gigapixel-scale Whole Slide Images (WSIs). However, recent methods leveraging guidance from Vision-Language Models often rely on static and universal pathological descriptions. This one-size-fits-all strategy fails to account for the vast morphological heterogeneity within individual WSIs, as its uniform guidance is not tailored to slide-specific visual evidence. To address this, we propose DyKo, a Dynamic Knowledge-guided MIL framework that adapts universal knowledge to slide-specific evidence for few-shot WSI classification. The core of DyKo is the WSI-Adaptive Knowledge Instantiation module (WAKI). WAKI begins by identifying key visual prototypes within a specific WSI's histology. These slide-specific prototypes then serve as queries to retrieve relevant concepts from a pathology knowledge base. This retrieved knowledge is then used to synthesize unique, knowledge-instantiated features for each instance, effectively instantiating tailored guidance at the patch level. To ensure fidelity and prevent semantic drift, we introduce a Structural Consistency loss that enforces alignment between knowledge-instantiated and visual features. Comprehensive experiments on four public real-world cancer datasets demonstrate that DyKo achieves superior performance over state-of-the-art methods in few-shot pathology diagnosis. Code is available at https://github.com/junjianli106/DyKo.

Monocular 3D object detection (Mono3D) aims to infer object locations and dimensions in 3D space from a single RGB image. Despite recent progress, existing methods remain highly sensitive to camera intrinsics and struggle to generalize across diverse settings, since intrinsic governs how 3D scenes are projected onto the image plane. We propose MonoIA, a unified intrinsic-aware framework that models and adapts to intrinsic variation through a language-grounded representation.The key insight is that intrinsic variation is not a numeric difference but a perceptual transformation that alters apparent scale, perspective, and spatial geometry. To capture this effect, MonoIA employs large language models and vision-language models to generate intrinsic embeddings that encode the visual and geometric implications of camera parameters.These embeddings are hierarchically integrated into the detection network via an Intrinsic Adaptation Module, allowing the model to modulate its feature representations according to camera-specific configurations and maintain consistent 3D detection across intrinsics.This shifts intrinsic modeling from numeric conditioning to semantic representation, enabling robust and unified perception across cameras.Extensive experiments show that MonoIA achieves new state-of-the-art results on standard benchmarks including KITTI, Waymo, and nuScenes (e.g., +1.18% on the KITTI leaderboard), and further improves performance under multi-dataset training (e.g., +4.46% on KITTI Val).

Generating animated 3D objects is at the heart of many applications, yet most advanced works are typically difficult to apply in practice because of their limited setup, their long runtime, or their limited quality. We introduce ActionMesh, a generative model that predicts production-ready 3D meshes "in action" in a feed-forward manner. Drawing inspiration from early video models, our key insight is to modify existing 3D diffusion models to include a temporal axis, resulting in a framework we dubbed "temporal 3D diffusion". Specifically, we first adapt the 3D diffusion stage to generate a sequence of synchronized latents representing time-varying and independent 3D shapes. Second, we design a temporal 3D autoencoder that translates a sequence of independent shapes into the corresponding deformations of a pre-defined reference shape, allowing us to build an animation. Combining these two components, ActionMesh generates animated 3D meshes from different inputs like a monocular video, a text description, or even a 3D mesh with a text prompt describing its animation. Besides, compared to previous approaches, our method is fast and produces results that are rig-free and topology consistent, hence enabling rapid iteration and seamless applications like texturing and retargeting. We evaluate our model on standard video-to-4D benchmarks (Consistent4D, Objaverse) and report state-of-the-art performances on both geometric accuracy and temporal consistency, demonstrating that our model can deliver animated 3D meshes with unprecedented speed and quality.

Avoiding catastrophic forgetting for previous tasks and maintaining model plasticity to support new tasks are two critical objectives of continual learning. However, existing methods usually neglect one of the two aspects and fail to support long task sequences with satisfactory performance, especially in resource-constrained scenarios in which the size of the model is limited. This work proposes GRAPA, a parameter-efficient continual learning method that well balances stability and plasticity of the model to handle long task sequences with diverse complexities. GRAPA enhances model plasticity without sacrificing stability with two novel designs. First, a gradient-guided parameter reuse strategy is proposed to make full use of frozen parameters while ensuring that no task interference is introduced. Second, a reinforcement-learning-based parameter allocation is designed to enable the model to adapt to the current task on top of reused parameters while preserving maximal model capacity for future tasks. Experiments on multiple task sequences composed of various datasets demonstrate that GRAPA lifts mean task accuracy by up to 7.67%, with up to 14.92% gains on subsequent complex tasks, reflecting GRAPA's superior plasticity.

We introduce WAM-Flow, a vision-language-action (VLA) model that casts ego-trajectory planning as discrete flow matching over a structured token space. In contrast to autoregressive decoders, WAM-Flow performs fully parallel, bidirectional denoising, enabling coarse-to-fine refinement with a tunable compute-accuracy trade-off. Specifically, the approach combines a metric-aligned numerical tokenizer that preserves scalar geometry via triplet-margin learning, a geometry-aware flow objective and a simulator-guided GRPO alignment that integrates safety, ego progress, and comfort rewards while retaining parallel generation. A multi-stage adaptation converts a pre-trained auto-regressive backbone (Janus-1.5B) from causal decoding to non-causal flow model and strengthens road-scene competence through continued multimodal pretraining. Thanks to the inherent nature of consistency model training and parallel decoding inference, WAM-Flow achieves superior closed-loop performance against autoregressive and diffusion-based VLA baselines, with 1-step inference attaining 88.7 PDMS and 5-step inference reaching 90.3 PDMS on NAVSIM v1 benchmark. These results establish discrete flow matching as a new promising paradigm for end-to-end autonomous driving.

In AI-generated image detection, current cutting-edge methods typically adapt pre-trained foundation models through partial-parameter fine-tuning. However, these approaches often struggle to generalize to forgeries from unseen generators, as the fine-tuned models capture only limited patterns from training data and fail to reflect the evolving traits of new ones. To overcome this limitation, we propose Image-Adaptive Prompt Learning (IAPL), a novel paradigm that dynamically adjusts the prompts fed into the encoder according to each testing image, rather than fixing them after training. This design significantly enhances robustness and adaptability to diverse forged images. The dynamic prompts integrate conditional information with test-time adaptive tokens through a lightweight learnable scaling factor. The conditional information is produced by a Conditional Information Learner, which leverages CNN-based feature extractors to model both forgery-specific and general conditions. The test-time adaptive tokens are optimized during inference on a single sample by enforcing prediction consistency across multiple views, ensuring that the parameters align with the current image. For the final decision, the optimal input with the highest prediction confidence is selected. Extensive experiments show that IAPL achieves state-of-the-art performance, with mean accuracies of 95.61% and 96.7% on the widely used UniversalFakeDetect and GenImage datasets, respectively.

Neural radiance fields (NeRF)-based methods with multi-resolution hash encoding enable efficient sparse-view CBCT reconstruction, but real-world projections violate ideal assumptions due to scatter/noise and related inconsistencies. Uniformly fusing hash-grid levels entangles heterogeneous frequency components, yielding spurious high-frequency textures in homogeneous tissues, blurred boundaries, and propagation of projection-induced bias. We propose GH-NAF (Grid-Adaptive Hash-Level-Attended Neural Attenuation Field), which decouples hash levels and adaptively weights them via uncertainty-guided, grid-adaptive hash-level attention. This stabilizes low-frequency modeling in homogeneous regions while selectively preserving high-frequency details near structural boundaries. Experiments on synthetic and real CBCT data show that GH-NAF improves intra-material contrast and reconstruction quality over state-of-the-art methods. The code is available at https://github.com/seongje-oh/GH-NAF