We empirically study autoregressive pre-training from videos. To perform our study, we construct a series of autoregressive video models, called Toto. We treat videos as sequences of visual tokens and train transformer models to autoregressively predict future tokens. Our models are pre-trained on a diverse dataset of videos and images comprising over 1 trillion visual tokens. We explore different architectural, training, and inference design choices. We evaluate the learned visual representations on a range of downstream tasks including image recognition, video classification, object tracking, and robotics. Our results demonstrate that, despite minimal inductive biases, autoregressive pre-training leads to competitive performance across all benchmarks. Finally, we find that scaling our video models results in similar scaling curves to those seen in language models, albeit with a different rate.
论文检索
输入标题、作者或关键词，从 12,319 篇学术成果中精准定位
While test-time adaptation (TTA) methods effectively address domain shifts by dynamically adapting pre-trained models to target domain data during online inference, their application to 3D point clouds is hindered by their irregular and unordered structure. Current 3D TTA methods often rely on computationally expensive spatial-domain optimizations and may require additional training data. In contrast, we propose Graph Spectral Domain Test-Time Adaptation (GSDTTA), a novel approach for 3D point cloud classification that shifts adaptation to the graph spectral domain, enabling more efficient adaptation by capturing global structural properties with fewer parameters. Point clouds in target domain are represented as outlier-aware graphs and transformed into graph spectral domain by Graph Fourier Transform (GFT). For efficiency, adaptation is performed by optimizing only the lowest 10% of frequency components, which capture the majority of the point cloud's energy. An inverse GFT (IGFT) is then applied to reconstruct the adapted point cloud with the graph spectral-driven point shift. This process is enhanced by an eigenmap-guided self-training strategy that iteratively refines both the spectral adjustments and the model parameters. Experimental results and ablation studies on benchmark datasets demonstrate the effectiveness of GSDTTA, outperforming existing TTA methods for 3D point cloud classification.
Spatiotemporal video grounding aims to localize target entities in videos based on textual queries. While existing research has made significant progress in exocentric videos, the egocentric setting remains relatively underexplored, despite its growing importance in applications such as augmented reality and robotics. In this work, we conduct a systematic analysis of the discrepancies between egocentric and exocentric videos, revealing key challenges such as shorter object durations, sparser trajectories, smaller object sizes, and larger positional shifts. To address these challenges, we introduce EgoMask, the first pixel-level benchmark for fine-grained spatiotemporal grounding in egocentric videos. It is constructed by our proposed automatic annotation pipeline, which annotates referring expressions and object masks across short-, medium-, and long-term videos. Additionally, we create EgoMask-Train, a large-scale training dataset to facilitate model development. Experiments demonstrate that the state-of-the-art spatiotemporal grounding models perform poorly on our benchmark EgoMask, but fine-tuning on EgoMask-Train yields significant improvements, while preserving performance on exocentric datasets. Our work thus provides essential resources and insights for advancing egocentric video understanding.
Estimating 2D camera motion is a fundamental computer vision task that models the projection of 3D camera movements onto the 2D image plane. Current methods rely on either homography-based approaches, limited to planar scenes, or meshflow techniques that use grid-based local homographies but struggle with complex non-linear transformations. We introduce CamFlow, a novel framework that represents camera motion using hybrid motion bases: physical bases derived from camera geometry and stochastic bases for complex scenarios. Our approach includes a hybrid probabilistic loss function based on the Laplace distribution that enhances training robustness. For evaluation, we create a new benchmark by masking dynamic objects in existing optical flow datasets to isolate pure camera motion. Experiments show CamFlow outperforms state-of-the-art methods across diverse scenarios, demonstrating superior robustness and generalization in zero-shot settings. Code and datasets are available at our project page: https://lhaippp.github.io/CamFlow/.
Human-centric generative models are becoming increasingly popular, giving rise to various innovative tools and applications, such as talking face videos conditioned on text or audio prompts. The core of these capabilities lies in powerful pre-trained foundation models, trained on large-scale, high-quality datasets. However, many advanced methods rely on in-house data subject to various constraints, and other current studies fail to generate high-resolution face videos, which is mainly attributed to the significant lack of large-scale, high-quality face video datasets. In this paper, we introduce a human face video dataset, DH-FaceVid-1K. Our collection spans 1,200 hours in total, encompassing 270,043 video clips from over 20,000 individuals. Each sample includes corresponding speech audio, facial keypoints, and text annotations. Compared to other publicly available datasets, ours distinguishes itself through its multi-ethnic coverage and high-quality, comprehensive individual attributes. We establish multiple face video generation models supporting tasks such as text-to-video and image-to-video generation. In addition, we develop comprehensive benchmarks to validate the scaling law when using different proportions of the proposed dataset. Our primary aim is to contribute a face video dataset, particularly addressing the underrepresentation of Asian faces in existing curated datasets and thereby enriching the global spectrum of face-centric data and mitigating demographic biases. Project Page: https://luna-ai-lab.github.io/DH-FaceVid-1K/.
Despite advances in general video understanding, Video Large Language Models (Video-LLMs) face challenges in precise temporal localization due to discrete time representations and limited temporally aware datasets. Existing methods for temporal expression either conflate time with text-based numerical values, add a series of dedicated temporal tokens, or regress time using specialized temporal grounding heads. To address these issues, we introduce DisTime, a lightweight framework designed to enhance temporal comprehension in Video-LLMs. DisTime employs a learnable token to create a continuous temporal embedding space and incorporates a Distribution-based Time Decoder that generates temporal probability distributions, effectively mitigating boundary ambiguities and maintaining temporal continuity. Additionally, the Distribution-based Time Encoder re-encodes timestamps to provide time markers for Video-LLMs. To overcome temporal granularity limitations in existing datasets, we propose an automated annotation paradigm that combines the captioning capabilities of Video-LLMs with the localization expertise of dedicated temporal models. This leads to the creation of InternVid-TG, a substantial dataset with 1.25M temporally grounded events across 179k videos, surpassing ActivityNet-Caption by 55 times. Extensive experiments demonstrate that DisTime achieves state-of-the-art performance across benchmarks in three time-sensitive tasks while maintaining competitive performance in Video QA tasks. DisTime is released at https://github.com/josephzpng/DisTime.
Most existing remote sensing instance segmentation approaches are designed for close-vocabulary prediction, limiting their ability to recognize novel categories or generalize across datasets. This restricts their applicability in diverse Earth observation scenarios. To address this, we introduce open-vocabulary (OV) learning for remote sensing instance segmentation. While current OV segmentation models perform well on natural image datasets, their direct application to remote sensing faces challenges such as diverse landscapes, seasonal variations, and the presence of small or ambiguous objects in aerial imagery. To overcome these challenges, we propose SCORE (Scene Context matters in Open-vocabulary REmote sensing instance segmentation), a framework that integrates multi-granularity scene context, i.e., regional context and global context, to enhance both visual and textual representations. Specifically, we introduce Region-Aware Integration, which refines class embeddings with regional context to improve object distinguishability. Additionally, we propose Global Context Adaptation, which enriches naive text embeddings with remote sensing global context, creating a more adaptable and expressive linguistic latent space for the classifier. We establish new benchmarks for OV remote sensing instance segmentation across diverse datasets. Experimental results demonstrate that, our proposed method achieves SOTA performance, which provides a robust solution for large-scale, real-world geospatial analysis. Our code is available at https://github.com/HuangShiqi128/SCORE.
Know Your Attention Maps: Class-specific Token Masking for Weakly Supervised Semantic Segmentation
PDF ↗Weakly Supervised Semantic Segmentation (WSSS) is a challenging problem that has been extensively studied in recent years. Traditional approaches often rely on external modules like Class Activation Maps to highlight regions of interest and generate pseudo segmentation masks. In this work, we propose an end-to-end method that directly utilizes the attention maps learned by a Vision Transformer (ViT) for WSSS. We propose training a sparse ViT with multiple [CLS] tokens (one for each class), using a random masking strategy to promote [CLS] token - class assignment. At inference time, we aggregate the different self-attention maps of each [CLS] token corresponding to the predicted labels to generate pseudo segmentation masks. Our proposed approach enhances the interpretability of self-attention maps and ensures accurate class assignments. Extensive experiments on two standard benchmarks and three specialized datasets demonstrate that our method generates accurate pseudo-masks, outperforming related works. Those pseudo-masks can be used to train a segmentation model which achieves results comparable to fully-supervised models, significantly reducing the need for fine-grained labeled data.
No published work on visual question answering (VQA) accounts for ambiguity regarding where the content described in the question is located in the image. To fill this gap, we introduce VQ-FocusAmbiguity, the first VQA dataset that visually grounds each plausible image region a question could refer to when arriving at valid answers. We next analyze and compare our dataset to existing datasets to reveal its unique properties. Finally, we benchmark modern models for two novel tasks related to acknowledging focus ambiguity: recognizing whether a visual question has focus ambiguity and locating all plausible focus regions within the image. Results show that the dataset is challenging for modern models. To facilitate future progress on these tasks, we publicly share the dataset with an evaluation server at https://vizwiz.org/tasks-and-datasets/focus-ambiguity-in-visual-questions/.
Are VLMs Ready for Autonomous Driving? An Empirical Study from the Reliability, Data and Metric Perspectives
PDF ↗Recent advancements in Vision-Language Models (VLMs) have fueled interest in autonomous driving applications, particularly for interpretable decision-making. However, the assumption that VLMs provide visually grounded and reliable driving explanations remains unexamined. To address this, we introduce DriveBench, a benchmark evaluating 12 VLMs across 17 settings, covering 19,200 images, 20,498 QA pairs, and four key driving tasks. Our findings reveal that VLMs often generate plausible responses from general knowledge or textual cues rather than true visual grounding, especially under degraded or missing visual inputs. This behavior, concealed by dataset imbalances and insufficient evaluation metrics, poses significant risks in safety-critical scenarios like autonomous driving. We further observe that VLMs possess inherent corruption-awareness but only explicitly acknowledge these issues when directly prompted. Given the challenges and inspired by the inherent corruption awareness, we propose Robust Agentic Utilization (RAU), leveraging VLMs' corruption awareness and agentic planning with external tools to enhance perception reliability for downstream tasks. Our study challenges existing evaluation paradigms and provides a roadmap toward more robust and interpretable autonomous driving systems.
MoMa-Kitchen: A 100K+ Benchmark for Affordance-Grounded Last-Mile Navigation in Mobile Manipulation
PDF ↗In mobile manipulation, navigation and manipulation are often treated as separate problems, resulting in a significant gap between merely approaching an object and engaging with it effectively. Many navigation approaches primarily define success by proximity to the target, often overlooking the necessity for optimal positioning that facilitates subsequent manipulation. To address this, we introduce \ours, a benchmark dataset comprising over 100k samples that provide training data for models to learn optimal final navigation positions for seamless transition to manipulation. Our dataset includes affordance-grounded floor labels collected from diverse kitchen environments, in which robotic mobile manipulators of different models attempt to grasp target objects amidst clutter. Using a fully automated pipeline, we simulate diverse real-world scenarios and generate affordance labels for optimal manipulation positions. Visual data are collected from RGB-D inputs captured by a first-person view camera mounted on the robotic arm, ensuring consistency in viewpoint during data collection. We also develop a lightweight baseline model, \ourmodel, for navigation affordance grounding that demonstrates promising performance on the \ours benchmark. Our approach enables models to learn affordance-based final positioning that accommodates different arm types and platform heights, thereby paving the way for more robust and generalizable integration of navigation and manipulation in embodied AI.
Establishing dense correspondences across image pairs is essential for tasks such as shape reconstruction and robot manipulation. In the challenging setting of matching across different categories, the function of an object, i.e., the effect that an object can cause on other objects, can guide how correspondences should be established. This is because object parts that enable specific functions often share similarities in shape and appearance. We derive the definition of dense functional correspondence based on this observation and propose a weakly-supervised learning paradigm to tackle the prediction task. The main insight behind our approach is that we can leverage vision-language models to pseudo-label multi-view images to obtain functional parts. We then integrate this with dense contrastive learning from pixel correspondences to distill both functional and spatial knowledge into a new model that can establish dense functional correspondence. Further, we curate synthetic and real evaluation datasets as task benchmarks. Our results demonstrate the advantages of our approach over baseline solutions consisting of off-the-shelf self-supervised image representations and grounded vision language models.
Perception is a fundamental task in the field of computer vision, encompassing a diverse set of subtasks that can be systematically categorized into four distinct groups based on two dimensions: prediction type and instruction type. Notably, existing researches often focus solely on a limited subset of these potential combinations, which constrains their applicability and versatility across various contexts. In response to this challenge, we present MVL-LM, a Multi-granular and Versatile Perception framework incorporating Visual Large Language Model. Our framework is designed to integrate both word-based and sentence-based perception tasks alongside box and mask predictions within a single architecture. MVL-LM features an innovative multi-granularity decoder in conjunction with a CoT-inspired dataset unification strategy, enabling seamless supervised fine-tuning across a wide spectrum of tasks, including but not limited to panoptic segmentation, detection, grounding, and referring expression segmentation. Furthermore, we introduce a query enhancement strategy aimed at harnessing the decoding and generative capabilities inherent in VLLMs. Extensive experiments conducted across a range of benchmarks in both word-based and sentence-based perception tasks substantiate the efficacy of our framework. The code will be available at https://github.com/xiangwentao666/MVP-LM.
Vision-language models (VLMs) are prone to object hal- lucinations, where they erroneously indicate the presence of certain objects in an image. Existing benchmarks quantify hallucinations using relatively small, labeled datasets. However, this approach is i) insufficient to assess hallucinations that arise in open-world settings, where VLMs are widely used, and ii) inadequate for detecting systematic errors in VLMs. We propose DASH (Detection and Assess- ment of Systematic Hallucinations), an automatic, large-scale pipeline designed to identify systematic hallucinations of VLMs on real-world images in an open-world setting. A key component is DASH-OPT for image-based retrieval, where we optimize over the "natural image manifold" to generate images that mislead the VLM. The output of DASH consists of clusters of real and semantically similar images for which the VLM hallucinates an object. We apply DASH to PaliGemma and two LLaVA-NeXT models across 380 object classes and, in total, find more than 19k clusters with 950k images. We study the transfer of the identified systematic hallucinations to other VLMs and show that fine-tuning PaliGemma with the model-specific images obtained with DASH mitigates object hallucinations.
Spatial intelligence (SI) represents a cognitive ability encompassing the visualization, manipulation, and reasoning about spatial relationships, underpinning disciplines from neuroscience to robotics. We introduce SITE, a benchmark dataset towards SI Thorough Evaluation in a standardized format of multi-choice visual question-answering, designed to assess large vision-language models' spatial intelligence across diverse visual modalities (single-image, multi-image, and video) and SI factors (figural to environmental scales, spatial visualization and orientation, intrinsic and extrinsic, static and dynamic). Our approach to curating the benchmark combines a bottom-up survey of existing datasets and a top-down strategy drawing upon three classification systems in cognitive science, which prompt us to design two novel types of tasks about view-taking and dynamic scenes. Extensive experiments reveal that leading models fall behind human experts, especially in spatial orientation, a fundamental SI factor. Moreover, we demonstrate a positive correlation between a model's spatial reasoning proficiency and its performance on an embodied AI task.
Although large visual-language models (LVLMs) have demonstrated strong performance in multimodal tasks, errors may occasionally arise due to biases during the reasoning process. Recently, reward models (RMs) have become increasingly pivotal in the reasoning process. Specifically, process RMs evaluate each reasoning step, outcome RMs focus on the assessment of reasoning results, and critique RMs perform error analysis on the entire reasoning process, followed by corrections. However, existing benchmarks for vision-language RMs (VLRMs) typically assess only a single aspect of their capabilities (e.g., distinguishing between two answers), thus limiting the all-round evaluation and restricting the development of RMs in the visual-language domain. To address this gap, we propose a comprehensive and challenging benchmark, dubbed as VLRMBench, encompassing 12,634 questions. VLRMBench is constructed based on three distinct types of datasets, covering mathematical reasoning, hallucination understanding, and multi-image understanding. We design 12 tasks across three major categories, focusing on evaluating VLRMs in the aspects of process understanding, outcome judgment, and critique generation. Extensive experiments are conducted on 21 open-source models and 5 advanced closed-source models, highlighting the challenges posed by VLRMBench. For instance, in the `Forecasting Future', a binary classification task, the advanced GPT-4o achieves only a 76.0% accuracy. The code is available at https://github.com/JCruan519/VLRMBench.
The pursuit of a generalizable stereo matching model, capable of performing well across varying resolutions and disparity ranges without dataset-specific fine-tuning, has revealed a fundamental trade-off. Iterative local search methods achieve high scores on constrained benchmarks, but their core mechanism inherently limits the global consistency required for true generalization. However, global matching architectures, while theoretically more robust, have historically been rendered infeasible by prohibitive computational and memory costs. We resolve this dilemma with S2M2: a global matching architecture that achieves state-of-the-art accuracy and high efficiency without relying on cost volume filtering or deep refinement stacks. Our design integrates a multi-resolution transformer for robust long-range correspondence, trained with a novel loss function that concentrates probability on feasible matches. This approach enables a more robust joint estimation of disparity, occlusion, and confidence. S2M2 establishes a new state of the art on Middlebury v3 and ETH3D benchmarks, significantly outperforming prior methods in most metrics while reconstructing high-quality details with competitive efficiency.
The rapid advancements in generative technology have emerged as a double-edged sword. While offering powerful tools that enhance convenience, they also pose significant social concerns. As defenders, current synthetic image detection methods often lack artifact-level textual interpretability and are overly focused on image manipulation detection, and current datasets usually suffer from outdated generators and a lack of fine-grained annotations. In this paper, we introduce SynthScars, a high-quality and diverse dataset consisting of 12,236 fully synthetic images with human-expert annotations. It features 4 distinct image content types, 3 categories of artifacts, and fine-grained annotations covering pixel-level segmentation, detailed textual explanations, and artifact category labels. Furthermore, we propose LEGION (LEarning to Ground and explain for Synthetic Image detectiON), a multimodal large language model (MLLM)-based image forgery analysis framework that integrates artifact detection, segmentation, and explanation. Building upon this capability, we further explore LEGION as a controller, integrating it into image refinement pipelines to guide the generation of higher-quality and more realistic images. Extensive experiments show that LEGION outperforms existing methods across multiple benchmarks, particularly surpassing the second-best traditional expert on SynthScars by 3.31% in mIoU and 7.75% in F1 score. Moreover, the refined images generated under its guidance exhibit stronger alignment with human preferences. More information about LEGION can be found at https://opendatalab.github.io/LEGION.
The Mamba model excels in anomaly detection through efficient long-range dependency modeling and linear complexity. However, Mamba-based anomaly detectors still face two critical challenges: (1) insufficient modeling of diverse local features leading to inaccurate detection of subtle anomalies; (2) spatial-wise scanning mechanism disrupting the spatial continuity of large-scale anomalies, resulting in incomplete localization. To address these challenges, we propose Wave-MambaAD, a wavelet-driven state space model for unified subtle and large-scale anomaly detection. Firstly, to capture subtle anomalies, we design a high-frequency state space model that employs horizontal, vertical, and diagonal scanning mechanisms for processing directionally aligned high-frequency components, enabling precise anomaly detection through multidimensional feature extraction. Secondly, for comprehensive localization of large-scale anomalies, we propose a low-frequency state space model implementing channel-adaptive dynamic scanning mechanisms to maintain structural coherence in global contexts, which facilitates large-scale anomaly detection via adaptive feature integration. Finally, we develop a dynamic spatial enhancement block to improve anomalous feature representation by enhancing feature diversity through coordinated inter-channel communication and adaptive gating mechanisms. Comprehensive experiments on benchmark anomaly detection datasets show that Wave-MambaAD achieves competitive performance at lower parameters and computational costs.
Vision-Language Models (VLMs) frequently hallucinate responses to visual queries, undermining their reliability for critical applications. However, quantifying the effect of such hallucinations in free-form responses to open-ended queries requires visually verifying each claim within the response, which is highly challenging. We propose Programmatic VLM Evaluation (PROVE), a new benchmarking paradigm for evaluating VLM responses to open-ended queries. To construct PROVE, we provide a large language model with a high-fidelity scene-graph representation constructed from a detailed image caption, and prompt it to generate i) diverse and challenging question-answer (QA) pairs that test a range of image understanding capabilities, and ii) programs that can be executed over the scene graph object to verify each QA pair. We thus construct a benchmark of 10.6k challenging but grounded visual QA pairs. Next, we propose a scene graph-based evaluation framework to programmatically measure both the helpfulness and truthfulness of a free-form model response without relying on subjective LLM judgments. We extensively benchmark a range of VLMs on PROVE, and uncover a concerning tradeoff where models that provide more helpful responses often hallucinate more, whereas truthful models tend to be less informative. PROVE serves as a foundation for developing next-generation VLMs that balance helpfulness with truthfulness. A snapshot of our dataset is available at https://prove-explorer-anon.netlify.app/.