Recent advances in video diffusion models have significantly improved character animation techniques. However, current approaches rely on basic structural conditions such as DWPose or SMPL-X to animate character images, limiting their effectiveness in open-domain scenarios with dynamic backgrounds or challenging human poses. In this paper, we introduce $\textbf{AniCrafter}$, a diffusion-based human-centric animation model that can seamlessly integrate and animate a given character into open-domain dynamic backgrounds while following given human motion sequences. Built on cutting-edge Image-to-Video (I2V) diffusion architectures, our model incorporates an innovative "avatar-background" conditioning mechanism that reframes open-domain human-centric animation as a restoration task, enabling more stable and versatile animation outputs. Experimental results demonstrate the superior performance of our method. Codes will be available at https://github.com/MyNiuuu/AniCrafter.
Benefiting from contrastively trained visual encoders on large-scale natural scene images, Large Multimodal Models (LMMs) have achieved remarkable performance across various visual perception tasks. However, the inherent limitations of contrastive learning upon summarized descriptions fundamentally restrict the capabilities of models in meticulous reasoning, particularly in crucial scenarios of geometric problem-solving. To enhance geometric understanding, we propose a novel hard negative contrastive learning framework for the vision encoder, which combines image-based contrastive learning using generation-based hard negatives created by perturbing diagram generation code, and text-based contrastive learning using rule-based negatives derived from modified geometric descriptions and retrieval-based negatives selected based on caption similarity. We train CLIP using our strong negative learning method, namely MMCLIP (Multimodal Math CLIP), and subsequently train an LMM for geometric problem-solving. Experiments show that our trained model, MMGeoLM, significantly outperforms other open-source models on three geometric reasoning benchmarks. Even with a size of 7B, it can rival powerful closed-source models like GPT-4o. We further study the impact of different negative sample construction methods and the number of negative samples on the geometric reasoning performance of LMM, yielding fruitful conclusions. The code and dataset are available at https://github.com/THU-KEG/MMGeoLM.
Recent advancements in radiance field rendering, exemplified by Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS), have significantly progressed 3D modeling and reconstruction. The use of multiple 360-degree omnidirectional images for these tasks is increasingly favored due to advantages in data acquisition and comprehensive scene capture. However, the inherent geometric distortions in common omnidirectional representations, such as equirectangular projection (particularly severe in polar regions and varying with latitude), pose substantial challenges to achieving high-fidelity 3D reconstructions. Current datasets, while valuable, often lack the specific focus, scene composition, and ground truth granularity required to systematically benchmark and drive progress in overcoming these omnidirectional-specific challenges. To address this critical gap, we introduce Omnidirectional Blender 3D (OB3D), a new synthetic dataset curated for advancing 3D reconstruction from multiple omnidirectional images. OB3D features diverse and complex 3D scenes generated from Blender 3D projects, with a deliberate emphasis on challenging scenarios. The dataset provides comprehensive ground truth, including omnidirectional RGB images, precise omnidirectional camera parameters, and pixel-aligned equirectangular maps for depth and normals, alongside evaluation metrics. By offering a controlled yet challenging environment, OB3Daims to facilitate the rigorous evaluation of existing methods and prompt the development of new techniques to enhance the accuracy and reliability of 3D reconstruction from omnidirectional images.
Gesture recognition is an important research area in the field of computer vision. Most gesture recognition efforts focus on close-set scenarios, thereby limiting the capacity to effectively handle unseen or novel gestures. We aim to address class-incremental gesture recognition, which entails the ability to accommodate new and previously unseen gestures over time. Specifically, we introduce a Prototype-Guided Pseudo Feature Replay (PGPFR) framework for data-free class-incremental gesture recognition. This framework comprises four components: Pseudo Feature Generation with Batch Prototypes (PFGBP), Variational Prototype Replay (VPR) for old classes, Truncated Cross-Entropy (TCE) for new classes, and Continual Classifier Re-Training (CCRT). To tackle the issue of catastrophic forgetting, the PFGBP dynamically generates a diversity of pseudo features in an online manner, leveraging class prototypes of old classes along with batch class prototypes of new classes. Furthermore, the VPR enforces consistency between the classifier's weights and the prototypes of old classes, leveraging class prototypes and covariance matrices to enhance robustness and generalization capabilities. The TCE mitigates the impact of domain differences of the classifier caused by pseudo features. Finally, the CCRT training strategy is designed to prevent overfitting to new classes and ensure the stability of features extracted from old classes. Extensive experiments conducted on two widely used gesture recognition datasets, namely SHREC 2017 3D and EgoGesture 3D, demonstrate that our approach outperforms existing state-of-the-art methods by 11.8\% and 12.8\% in terms of mean global accuracy, respectively. The code is available on https://github.com/sunao-101/PGPFR-3/.
Long-term Action Quality Assessment (AQA) aims to evaluate the quantitative performance of actions in long videos. However, existing methods face challenges due to domain shifts between the pre-trained large-scale action recognition backbones and the specific AQA task, thereby hindering their performance. This arises since fine-tuning resource-intensive backbones on small AQA datasets is impractical. We address this by identifying two levels of domain shift: task-level, regarding differences in task objectives, and feature-level, regarding differences in important features. For feature-level shifts, which are more detrimental, we propose Progressive Hierarchical Instruction (PHI) with two strategies. First, Gap Minimization Flow (GMF) leverages flow matching to progressively learn a fast flow path that reduces the domain gap between initial and desired features across shallow to deep layers. Additionally, a temporally-enhanced attention module captures long-range dependencies essential for AQA. Second, List-wise Contrastive Regularization (LCR) facilitates coarse-to-fine alignment by comprehensively comparing batch pairs to learn fine-grained cues while mitigating domain shift. Integrating these modules, PHI offers an effective solution. Experiments demonstrate that PHI achieves state-of-the-art performance on three representative long-term AQA datasets, proving its superiority in addressing the domain shift for long-term AQA.
Current video-based fall detection research mostly relies on small, staged datasets with significant domain biases concerning background, lighting, and camera setup resulting in unknown real-world performance. We introduce OmniFall, unifying eight public fall detection datasets (roughly 14 h of recordings, roughly 42 h of multiview data, 101 subjects, 29 camera views) under a consistent ten-class taxonomy with standardized evaluation protocols. Our benchmark provides complete video segmentation labels and enables fair cross-dataset comparison previously impossible with incompatible annotation schemes. For real-world evaluation we curate OOPS-Fall from genuine accident videos and establish a staged-to-wild protocol measuring generalization from controlled to uncontrolled environments. Experiments with frozen pre-trained backbones such as I3D or VideoMAE reveal significant performance gaps between in-distribution and in-the-wild scenarios, highlighting critical challenges in developing robust fall detection systems. OmniFall Dataset at https://huggingface.co/datasets/simplexsigil2/omnifall , Code at https://github.com/simplexsigil/omnifall-experiments
Recent advancements in reasoning capability of Multimodal Large Language Models (MLLMs) demonstrate its effectiveness in tackling complex visual tasks. However, existing MLLM-based Video Anomaly Detection (VAD) methods remain limited to shallow anomaly descriptions without deep reasoning. In this paper, we propose a new task named Video Anomaly Reasoning (VAR), which aims to enable deep analysis and understanding of anomalies in the video by requiring MLLMs to think explicitly before answering. To this end, we propose Vad-R1, an end-to-end MLLM-based framework for VAR. Specifically, we design a Perception-to-Cognition Chain-of-Thought (P2C-CoT) that simulates the human process of recognizing anomalies, guiding the MLLM to reason anomaly step-by-step. Based on the structured P2C-CoT, we construct Vad-Reasoning, a dedicated dataset for VAR. Furthermore, we propose an improved reinforcement learning algorithm AVA-GRPO, which explicitly incentivizes the anomaly reasoning capability of MLLMs through a self-verification mechanism with limited annotations. Experimental results demonstrate that Vad-R1 achieves superior performance, outperforming both open-source and proprietary models on VAD and VAR tasks. Codes and datasets will be released at https://github.com/wbfwonderful/Vad-R1.
We introduce FruitNeRF++, a novel fruit-counting approach that combines contrastive learning with neural radiance fields to count fruits from unstructured input photographs of orchards. Our work is based on FruitNeRF, which employs a neural semantic field combined with a fruit-specific clustering approach. The requirement for adaptation for each fruit type limits the applicability of the method, and makes it difficult to use in practice. To lift this limitation, we design a shape-agnostic multi-fruit counting framework, that complements the RGB and semantic data with instance masks predicted by a vision foundation model. The masks are used to encode the identity of each fruit as instance embeddings into a neural instance field. By volumetrically sampling the neural fields, we extract a point cloud embedded with the instance features, which can be clustered in a fruit-agnostic manner to obtain the fruit count. We evaluate our approach using a synthetic dataset containing apples, plums, lemons, pears, peaches, and mangoes, as well as a real-world benchmark apple dataset. Our results demonstrate that FruitNeRF++ is easier to control and compares favorably to other state-of-the-art methods.
Neural Radiance Fields (NeRF) have transformed novel view synthesis by modeling scene-specific volumetric representations directly from images. While generalizable NeRF models can generate novel views across unknown scenes by learning latent ray representations, their performance heavily depends on a large number of multi-view observations. However, with limited input views, these methods experience significant degradation in rendering quality. To address this limitation, we propose GoLF-NRT: a Global and Local feature Fusion-based Neural Rendering Transformer. GoLF-NRT enhances generalizable neural rendering from few input views by leveraging a 3D transformer with efficient sparse attention to capture global scene context. In parallel, it integrates local geometric features extracted along the epipolar line, enabling high-quality scene reconstruction from as few as 1 to 3 input views. Furthermore, we introduce an adaptive sampling strategy based on attention weights and kernel regression, improving the accuracy of transformer-based neural rendering. Extensive experiments on public datasets show that GoLF-NRT achieves state-of-the-art performance across varying numbers of input views, highlighting the effectiveness and superiority of our approach. Code is available at https://github.com/KLMAV-CUC/GoLF-NRT.
Traditional approaches to adapting multi-modal large language models (MLLMs) to new tasks have relied heavily on fine-tuning. This paper introduces Efficient Multi-Modal Long Context Learning (EMLoC), a novel training-free alternative that embeds demonstration examples directly into the model input. EMLoC offers a more efficient, flexible, and scalable solution for task adaptation. Because extremely lengthy inputs introduce prohibitive computational and memory overhead, EMLoC contributes a chunk-wise compression mechanism combined with layer-wise adaptive pruning. It condenses long-context multimodal inputs into compact, task-specific memory representations. By adaptively pruning tokens at each layer under a Jensen-Shannon divergence constraint, our method achieves a dramatic reduction in inference complexity without sacrificing performance. This approach is the first to seamlessly integrate compression and pruning techniques for multi-modal long-context learning, offering a scalable and efficient solution for real-world applications. Extensive experiments on diverse vision-language benchmarks demonstrate that EMLoC achieves performance on par with or superior to naive long-context approaches. Our results highlight the potential of EMLoC as a groundbreaking framework for efficient and flexible adaptation of multi-modal models in resource-constrained environments. Codes are publicly available at https://github.com/Zehong-Ma/EMLoC.
The design of convolutional neural architectures that are exactly equivariant to continuous translations is an active field of research. It promises to benefit scientific computing, notably by making existing imaging systems more physically accurate. Most efforts focus on the design of downsampling/pooling layers, upsampling layers and activation functions, but little attention is dedicated to normalization layers. In this work, we present a novel theoretical framework for understanding the equivariance of normalization layers to discrete shifts and continuous translations. We also determine necessary and sufficient conditions for normalization layers to be equivariant in terms of the dimensions they operate on. Using real feature maps from ResNet-18 and ImageNet, we test those theoretical results empirically and find that they are consistent with our predictions.
Equivariant and invariant deep learning models have been developed to exploit intrinsic symmetries in data, demonstrating significant effectiveness in certain scenarios. However, these methods often suffer from limited representation accuracy and rely on strict symmetry assumptions that may not hold in practice. These limitations pose a significant drawback for image restoration tasks, which demands high accuracy and precise symmetry representation. To address these challenges, we propose a rotation-equivariant regularization strategy that adaptively enforces the appropriate symmetry constraints on the data while preserving the network's representational accuracy. Specifically, we introduce EQ-Reg, a regularizer designed to enhance rotation equivariance, which innovatively extends the insights of data-augmentation-based and equivariant-based methodologies. This is achieved through self-supervised learning and the spatial rotation and cyclic channel shift of feature maps deduce in the equivariant framework. Our approach firstly enables a non-strictly equivariant network suitable for image restoration, providing a simple and adaptive mechanism for adjusting equivariance based on task. Extensive experiments across three low-level tasks demonstrate the superior accuracy and generalization capability of our method, outperforming state-of-the-art approaches.
Recent advancements in generalizable novel view synthesis have achieved impressive quality through interpolation between nearby views. However, rendering high-resolution images remains computationally intensive due to the need for dense sampling of all rays. Recognizing that natural scenes are typically piecewise smooth and sampling all rays is often redundant, we propose a novel depth-guided bundle sampling strategy to accelerate rendering. By grouping adjacent rays into a bundle and sampling them collectively, a shared representation is generated for decoding all rays within the bundle. To further optimize efficiency, our adaptive sampling strategy dynamically allocates samples based on depth confidence, concentrating more samples in complex regions while reducing them in smoother areas. When applied to ENeRF, our method achieves up to a 1.27 dB PSNR improvement and a 47% increase in FPS on the DTU dataset. Extensive experiments on synthetic and real-world datasets demonstrate state-of-the-art rendering quality and up to 2x faster rendering compared to existing generalizable methods. Code is available at https://github.com/KLMAV-CUC/GDB-NeRF.
Human-centered images often suffer from severe generic degradation during transmission and are prone to human motion blur (HMB), making restoration challenging. Existing research lacks sufficient focus on these issues, as both problems often coexist in practice. To address this, we design a degradation pipeline that simulates the coexistence of HMB and generic noise, generating synthetic degraded data to train our proposed HAODiff, a human-aware one-step diffusion. Specifically, we propose a triple-branch dual-prompt guidance (DPG), which leverages high-quality images, residual noise (LQ minus HQ), and HMB segmentation masks as training targets. It produces a positive-negative prompt pair for classifier-free guidance (CFG) in a single diffusion step. The resulting adaptive dual prompts let HAODiff exploit CFG more effectively, boosting robustness against diverse degradations. For fair evaluation, we introduce MPII-Test, a benchmark rich in combined noise and HMB cases. Extensive experiments show that our HAODiff surpasses existing state-of-the-art (SOTA) methods in terms of both quantitative metrics and visual quality on synthetic and real-world datasets, including our introduced MPII-Test. Code is available at: https://github.com/gobunu/HAODiff.
Camera sensor simulation serves as a critical role for autonomous driving (AD), e.g. evaluating vision-based AD algorithms. While existing approaches have leveraged generative models for controllable image/video generation, they remain constrained to generating multi-view video sequences with fixed camera viewpoints and video frequency, significantly limiting their downstream applications. To address this, we present a generalizable camera simulation framework DriveCamSim, whose core innovation lies in the proposed Explicit Camera Modeling (ECM) mechanism. Instead of implicit interaction through vanilla attention, ECM establishes explicit pixel-wise correspondences across multi-view and multi-frame dimensions, decoupling the model from overfitting to the specific camera configurations (intrinsic/extrinsic parameters, number of views) and temporal sampling rates presented in the training data. For controllable generation, we identify the issue of information loss inherent in existing conditional encoding and injection pipelines, proposing an information-preserving control mechanism. This control mechanism not only improves conditional controllability, but also can be extended to be identity-aware to enhance temporal consistency in foreground object rendering. With above designs, our model demonstrates superior performance in both visual quality and controllability, as well as generalization capability across spatial-level (camera parameters variations) and temporal-level (video frame rate variations), enabling flexible user-customizable camera simulation tailored to diverse application scenarios. Code will be avaliable at https://github.com/swc-17/DriveCamSim for facilitating future research.
Vision-Language Models (VLMs) excel across diverse tasks but suffer from high inference costs in time and memory. Token sparsity mitigates inefficiencies in token usage, while neuron sparsity reduces high-dimensional computations, both offering promising solutions to enhance efficiency. Recently, these two sparsity paradigms have evolved largely in parallel, fostering the prevailing assumption that they function independently. However, a fundamental yet underexplored question remains: Do they truly operate in isolation, or is there a deeper underlying interplay that has yet to be uncovered? In this paper, we conduct the first comprehensive investigation into this question. By introducing and analyzing the matching mechanism between Core Neurons and Core Tokens, we found that key neurons and tokens for inference mutually influence and reinforce each other. Building on this insight, we propose CoreMatching, a co-adaptive sparse inference framework, which leverages the synergy between token and neuron sparsity to enhance inference efficiency. Through theoretical analysis and efficiency evaluations, we demonstrate that the proposed method surpasses state-of-the-art baselines on ten image understanding tasks and three hardware devices. Notably, on the NVIDIA Titan Xp, it achieved 5x FLOPs reduction and a 10x overall speedup. Code is released at https://github.com/wangqinsi1/2025-ICML-CoreMatching/tree/main.
Advanced autoregressive models have reshaped multimodal AI. However, their transformative potential in medical imaging remains largely untapped due to the absence of a unified visual tokenizer -- one capable of capturing fine-grained visual structures for faithful image reconstruction and realistic image synthesis, as well as rich semantics for accurate diagnosis and image interpretation. To this end, we present MedITok, the first unified tokenizer tailored for medical images, encoding both low-level structural details and high-level clinical semantics within a unified latent space. To balance these competing objectives, we introduce a novel two-stage training framework: a visual representation alignment stage that cold-starts the tokenizer reconstruction learning with a visual semantic constraint, followed by a textual semantic representation alignment stage that infuses detailed clinical semantics into the latent space. Trained on the meticulously collected large-scale dataset with over 30 million medical images and 2 million image-caption pairs, MedITok achieves state-of-the-art performance on more than 30 datasets across 9 imaging modalities and 4 different tasks. By providing a unified token space for autoregressive modeling, MedITok supports a wide range of tasks in clinical diagnostics and generative healthcare applications. Model and code will be made publicly available at: https://github.com/Masaaki-75/meditok.
Modality fusion is a cornerstone of multimodal learning, enabling information integration from diverse data sources. However, vanilla fusion methods are limited by (1) inability to account for heterogeneous interactions between modalities and (2) lack of interpretability in uncovering the multimodal interactions inherent in the data. To this end, we propose I2MoE (Interpretable Multimodal Interaction-aware Mixture of Experts), an end-to-end MoE framework designed to enhance modality fusion by explicitly modeling diverse multimodal interactions, as well as providing interpretation on a local and global level. First, I2MoE utilizes different interaction experts with weakly supervised interaction losses to learn multimodal interactions in a data-driven way. Second, I2MoE deploys a reweighting model that assigns importance scores for the output of each interaction expert, which offers sample-level and dataset-level interpretation. Extensive evaluation of medical and general multimodal datasets shows that I2MoE is flexible enough to be combined with different fusion techniques, consistently improves task performance, and provides interpretation across various real-world scenarios. Code is available at https://github.com/Raina-Xin/I2MoE.
The rapid advancement of large language models (LLMs) and multi-modal LLMs (MLLMs) has historically relied on model-centric scaling through increasing parameter counts from millions to hundreds of billions to drive performance gains. However, as we approach hardware limits on model size, the dominant computational bottleneck has fundamentally shifted to the quadratic cost of self-attention over long token sequences, now driven by ultra-long text contexts, high-resolution images, and extended videos. In this position paper, \textbf{we argue that the focus of research for efficient AI is shifting from model-centric compression to data-centric compression}. We position token compression as the new frontier, which improves AI efficiency via reducing the number of tokens during model training or inference. Through comprehensive analysis, we first examine recent developments in long-context AI across various domains and establish a unified mathematical framework for existing model efficiency strategies, demonstrating why token compression represents a crucial paradigm shift in addressing long-context overhead. Subsequently, we systematically review the research landscape of token compression, analyzing its fundamental benefits and identifying its compelling advantages across diverse scenarios. Furthermore, we provide an in-depth analysis of current challenges in token compression research and outline promising future directions. Ultimately, our work aims to offer a fresh perspective on AI efficiency, synthesize existing research, and catalyze innovative developments to address the challenges that increasing context lengths pose to the AI community's advancement.
Recent advancements in Large Vision-Language Models (VLMs), have greatly enhanced their capability to jointly process text and images. However, despite extensive benchmarks evaluating visual comprehension (e.g., diagrams, color schemes, OCR tasks...), there is limited assessment of VLMs' ability to read and reason about text-rich images effectively. To fill this gap, we introduce ReadBench, a multimodal benchmark specifically designed to evaluate the reading comprehension capabilities of VLMs. ReadBench transposes contexts from established text-only benchmarks into images of text while keeping textual prompts and questions intact. Evaluating leading VLMs with ReadBench, we find minimal-but-present performance degradation on short, text-image inputs, while performance sharply declines for longer, multi-page contexts. Our experiments further reveal that text resolution has negligible effects on multimodal performance. These findings highlight needed improvements in VLMs, particularly their reasoning over visually presented extensive textual content, a capability critical for practical applications. ReadBench is available at https://github.com/answerdotai/ReadBench .