We introduce PerceptionComp, a manually annotated benchmark for complex, long-horizon, perception-centric video reasoning. PerceptionComp is designed so that no single moment is sufficient: answering each question requires multiple temporally separated pieces of visual evidence and compositional constraints under conjunctive and sequential logic, spanning perceptual subtasks such as objects, attributes, relations, locations, actions, and events, and requiring skills including semantic recognition, visual correspondence, temporal reasoning, and spatial reasoning. The benchmark contains 1,114 highly complex questions on 279 videos from diverse domains including city walk tours, indoor villa tours, video games, and extreme outdoor sports, with 100

The integration of visual understanding and generation into unified multimodal models represents a significant stride toward general-purpose AI. However, a fundamental question remains unanswered by existing benchmarks: does this architectural unification actually enable synergetic interaction between the constituent capabilities? Existing evaluation paradigms, which primarily assess understanding and generation in isolation, are insufficient for determining whether a unified model can leverage its understanding to enhance its generation, or use generative simulation to facilitate deeper comprehension. To address this critical gap, we introduce RealUnify, a benchmark specifically designed to evaluate bidirectional capability synergy. RealUnify comprises 1,000 meticulously human-annotated instances spanning 10 categories and 32 subtasks. It is structured around two core axes: 1) Understanding Enhances Generation, which requires reasoning (e.g., commonsense, logic) to guide image generation, and 2) Generation Enhances Understanding, which necessitates mental simulation or reconstruction (e.g., of transformed or disordered visual inputs) to solve reasoning tasks. A key contribution is our dual-evaluation protocol, which combines direct end-to-end assessment with a diagnostic stepwise evaluation that decomposes tasks into distinct understanding and generation phases. This protocol allows us to precisely discern whether performance bottlenecks stem from deficiencies in core abilities or from a failure to integrate them. Through large-scale evaluations of 12 leading unified models and 6 specialized baselines, we find that current unified models still struggle to achieve effective synergy, indicating that architectural unification alone is insufficient. These results highlight the need for new training strategies and inductive biases to fully unlock the potential of unified modeling.

Large Language Models (LLMs) have achieved remarkable reliability and advanced capabilities through extended test-time reasoning. However, extending these capabilities to Multi-modal Large Language Models (MLLMs) remains a significant challenge due to a critical scarcity of high-quality, long-chain reasoning data and optimized training pipelines. To bridge this gap, we present a unified multi-agent visual reasoning framework that systematically evolves from our foundational image-centric model, Insight-V, into a generalized spatial-temporal architecture, Insight-V++. We first propose a scalable data generation pipeline equipped with multi-granularity assessment that autonomously synthesizes structured, complex reasoning trajectories across image and video domains without human intervention. Recognizing that directly supervising MLLMs with such intricate data yields sub-optimal results, we design a dual-agent architecture comprising a reasoning agent to execute extensive analytical chains, and a summary agent to critically evaluate and distill final outcomes. While our initial framework utilized Direct Preference Optimization (DPO), its off-policy nature fundamentally constrained reinforcement learning potential. To overcome these limitations, particularly for long-horizon video understanding, Insight-V++ introduces two novel algorithms, ST-GRPO and J-GRPO, which enhance spatial-temporal reasoning and improve evaluative robustness. Crucially, by leveraging reliable feedback from the summary agent, we guide an iterative reasoning path generation process, retraining the entire multi-agent system in a continuous, self-improving loop. Extensive experiments on base models like LLaVA-NeXT and Qwen2.5-VL demonstrate significant performance gains across challenging image and video reasoning benchmarks while preserving strong capabilities on traditional perception-focused tasks.

Embodied Vision-Language Models (VLMs) have demonstrated impressive performance and generalization in robotics, particularly within Vision-Language-Action frameworks. However, a significant gap remains between the high-level semantic focus of standard text-guided pre-training paradigms and the low-level spatial and physical knowledge critical for execution in embodied environments. In this paper, we introduce GEM, a Generative-supervised Embodied vision-language Model designed to bridge this divide. We propose integrating a depth map generation task directly into the VLM pre-training phase. By training this generative objective jointly with the main model, we observe substantial improvements in embodied intelligence, significantly enhancing both semantic understanding and physical operation capabilities. To support this paradigm, we curate and release GEM-4M, a comprehensive large-scale dataset featuring a mixture of grounding, reasoning, and planning data paired with high-quality depth supervision. Extensive experiments demonstrate that GEM achieves state-of-the-art results across diverse embodied benchmarks. Furthermore, our deployed action model, GEM-VLA, exhibits vastly superior task execution abilities in both simulation environments and real-world evaluations. Code, models, and datasets are available at https://zhaorw02.github.io/GEM/

A unified representation for text and vision is a natural pursuit, as it enables simpler multimodal modeling and more efficient training. However, representing images as discrete signals in the same way as text inevitably introduces severe information loss. Existing work struggles to balance low-level details and high-level semantics in discrete representations: reconstruction-oriented representations often lack semantic information, whereas semantically stronger features typically suffer from severe loss of detail. We present ViQ, a Visual Quantized Representations framework, which is designed to balance semantics and details in discrete representations while supporting inputs at native resolutions, thereby enabling it to serve as a unified and general discrete representation for arbitrary visual inputs. Our approach structures quantization learning into two stages: text-aligned pre-training and feature discretization. With text-aligned pre-training, we enhance the visual encoder semantic-rich supervision from the pretrained language model and enable it to process native-resolution visual inputs. During discretization, we propose a proximal representation learning strategy to progressively compact the feature space, along with a position-aware head-wise quantization mechanism that enables flexible processing of arbitrary resolutions. Extensive experiments on multimodal tasks demonstrate that ViQ achieves competitive performance compared to state-of-the-art multimodal vision encoders with continuous and high-dimensional visual features, while maintaining high precision in low-level reconstruction. We also show that multimodal training with visual quantized representations largely improves efficiency, yielding up to 20%-70% acceleration with different base LLMs and training recipes.

Recently, we have witnessed the great success of the generalist model in natural language processing. The generalist model is a general framework trained with massive data and is able to process various downstream tasks simultaneously. Encouraged by their impressive performance, an increasing number of researchers are venturing into the realm of applying these models to computer vision tasks. However, the inputs and outputs of vision tasks are more diverse, and it is difficult to summarize them as a unified representation. In this paper, we provide a comprehensive overview of the vision generalist models, delving into their characteristics and capabilities within the field. First, we review the background, including the datasets, tasks, and benchmarks. Then, we dig into the design of frameworks that have been proposed in existing research, while also introducing the techniques employed to enhance their performance. To better help the researchers comprehend the area, we take a brief excursion into related domains, shedding light on their interconnections and potential synergies. To conclude, we provide some real-world application scenarios, undertake a thorough examination of the persistent challenges, and offer insights into possible directions for future research endeavors.

Recent advances in large language models, particularly following GPT-4o, have sparked increasing interest in developing omni-modal models capable of understanding more modalities. While some open-source alternatives have emerged, there is still a notable lag behind specialized single-modality models in performance. In this paper, we present Ola, an Omni-modal Language model that achieves competitive performance across image, video, and audio understanding compared to specialized counterparts, pushing the frontiers of the omni-modal language model to a large extent. We conduct a comprehensive exploration of architectural design, data curation, and training strategies essential for building a robust omni-modal model. Ola incorporates advanced visual understanding and audio recognition capabilities through several critical and effective improvements over mainstream baselines. Moreover, we rethink inter-modal relationships during omni-modal training, emphasizing cross-modal alignment with video as a central bridge, and propose a progressive training pipeline that begins with the most distinct modalities and gradually moves towards closer modality alignment. Extensive experiments demonstrate that Ola surpasses existing open omni-modal LLMs across all modalities while achieving highly competitive performance compared to state-of-the-art specialized models of similar sizes. We aim to make Ola a fully open omni-modal understanding solution to advance future research in this emerging field. Model weights, code, and data are open-sourced at https://github.com/Ola-Omni/Ola.

Large Language Models (LLMs) demonstrate enhanced capabilities and reliability by reasoning more, evolving from Chain-of-Thought prompting to product-level solutions like OpenAI o1. Despite various efforts to improve LLM reasoning, high-quality long-chain reasoning data and optimized training pipelines still remain inadequately explored in vision-language tasks. In this paper, we present Insight-V, an early effort to 1) scalably produce long and robust reasoning data for complex multi-modal tasks, and 2) an effective training pipeline to enhance the reasoning capabilities of multi-modal large language models (MLLMs). Specifically, to create long and structured reasoning data without human labor, we design a two-step pipeline with a progressive strategy to generate sufficiently long and diverse reasoning paths and a multi-granularity assessment method to ensure data quality. We observe that directly supervising MLLMs with such long and complex reasoning data will not yield ideal reasoning ability. To tackle this problem, we design a multi-agent system consisting of a reasoning agent dedicated to performing long-chain reasoning and a summary agent trained to judge and summarize reasoning results. We further incorporate an iterative DPO algorithm to enhance the reasoning agent's generation stability and quality. Based on the popular LLaVA-NeXT model and our stronger base MLLM, we demonstrate significant performance gains across challenging multi-modal benchmarks requiring visual reasoning. Benefiting from our multi-agent system, Insight-V can also easily maintain or improve performance on perception-focused multi-modal tasks.

Current research on agentic visual reasoning enables deep multimodal understanding but primarily focuses on image manipulation tools, leaving a gap toward more general-purpose agentic models. In this work, we revisit the geolocalization task, which requires not only nuanced visual grounding but also web search to confirm or refine hypotheses during reasoning. Since existing geolocalization benchmarks fail to meet the need for high-resolution imagery and the localization challenge for deep agentic reasoning, we curate GeoBench, a benchmark that includes photos and panoramas from around the world, along with a subset of satellite images of different cities to rigorously evaluate the geolocalization ability of agentic models. We also propose GeoVista, an agentic model that seamlessly integrates tool invocation within the reasoning loop, including an image-zoom-in tool to magnify regions of interest and a web-search tool to retrieve related web information. We develop a complete training pipeline for it, including a cold-start supervised fine-tuning (SFT) stage to learn reasoning patterns and tool-use priors, followed by a reinforcement learning (RL) stage to further enhance reasoning ability. We adopt a hierarchical reward to leverage multi-level geographical information and improve overall geolocalization performance. Experimental results show that GeoVista surpasses other open-source agentic models on the geolocalization task greatly and achieves performance comparable to closed-source models such as Gemini-2.5-flash and GPT-5 on most metrics.

In the realm of vision-language understanding, the proficiency of models in interpreting and reasoning over visual content has become a cornerstone for numerous applications. However, it is challenging for the visual encoder in Large Vision-Language Models (LVLMs) to extract useful features tailored to questions that aid the language model's response. Furthermore, a common practice among existing LVLMs is to utilize lower-resolution images, which restricts the ability for visual recognition. Our work introduces the Chain-of-Spot (CoS) method, which we describe as Interactive Reasoning, a novel approach that enhances feature extraction by focusing on key regions of interest (ROI) within the image, corresponding to the posed questions or instructions. This technique allows LVLMs to access more detailed visual information without altering the original image resolution, thereby offering multi-granularity image features. By integrating Chain-of-Spot with instruct-following LLaVA-1.5 models, the process of image reasoning consistently improves performance across a wide range of multimodal datasets and benchmarks without bells and whistles and achieves new state-of-the-art results. Our empirical findings demonstrate a significant improvement in LVLMs' ability to understand and reason about visual content, paving the way for more sophisticated visual instruction-following applications. Code and models are available at https://github.com/dongyh20/Chain-of-Spot

In the field of instruction-following large vision-language models (LVLMs), the efficient deployment of these models faces challenges, notably due to the high memory demands of their key-value (KV) caches. Conventional cache management strategies for LLMs focus on cache eviction, which often fails to address the specific needs of multi-modal instruction-following models. Recognizing this gap, in this paper, we introduce Elastic Cache, a novel approach that benefits from applying distinct acceleration methods for instruction encoding and output generation stages. We investigate the metrics of importance in different stages and propose an 'importance-driven cache merging' strategy to prune redundancy caches. Instead of discarding less important caches, our strategy identifies important key/value vectors as anchor points. Surrounding less important caches are then merged with these anchors, enhancing the preservation of contextual information in the KV caches while yielding an arbitrary acceleration ratio. For instruction encoding, we utilize the frequency to evaluate the importance of caches. Regarding output generation, we prioritize tokens based on their 'distance' with an offset, by which both the initial and most recent tokens are retained. Results on a range of LVLMs demonstrate that Elastic Cache not only boosts efficiency but also notably outperforms existing pruning methods in language generation across various tasks. Code is available at https://github.com/liuzuyan/ElasticCache.

Diffusion models (DMs) have become the new trend of generative models and have demonstrated a powerful ability of conditional synthesis. Among those, text-to-image diffusion models pre-trained on large-scale image-text pairs are highly controllable by customizable prompts. Unlike the unconditional generative models that focus on low-level attributes and details, text-to-image diffusion models contain more high-level knowledge thanks to the vision-language pre-training. In this paper, we propose VPD (Visual Perception with pre-trained Diffusion models), a new framework that exploits the semantic information of a pre-trained text-to-image diffusion model in visual perception tasks. Instead of using the pre-trained denoising autoencoder in a diffusion-based pipeline, we simply use it as a backbone and aim to study how to take full advantage of the learned knowledge. Specifically, we prompt the denoising decoder with proper textual inputs and refine the text features with an adapter, leading to a better alignment to the pre-trained stage and making the visual contents interact with the text prompts. We also propose to utilize the cross-attention maps between the visual features and the text features to provide explicit guidance. Compared with other pre-training methods, we show that vision-language pre-trained diffusion models can be faster adapted to downstream visual perception tasks using the proposed VPD. Extensive experiments on semantic segmentation, referring image segmentation, and depth estimation demonstrate the effectiveness of our method. Notably, VPD attains 0.254 RMSE on NYUv2 depth estimation and 73.3% oIoU on RefCOCO-val referring image segmentation, establishing new records on these two benchmarks. Code is available at https://github.com/wl-zhao/VPD.

In this paper, we propose DiffSwap, a diffusion model based framework for high-fidelity and controllable face swapping. Unlike previous work that relies on carefully designed network architectures and loss functions to fuse the information from the source and target faces, we reformulate the face swapping as a conditional inpainting task, performed by a powerful diffusion model guided by the desired face attributes (e.g., identity and landmarks). An important issue that makes it nontrivial to apply diffusion models to face swapping is that we cannot perform the time-consuming multi-step sampling to obtain the generated image during training. To overcome this, we propose a midpoint estimation method to efficiently recover a reasonable diffusion result of the swapped face with only 2 steps, which enables us to introduce identity constraints to improve the face swapping quality. Our framework enjoys several favorable properties more appealing than prior arts: 1) Controllable. Our method is based on conditional masked diffusion on the latent space, where the mask and the conditions can be fully controlled and customized. 2) High-fidelity. The formulation of conditional inpainting can fully exploit the generative ability of diffusion models and can preserve the background of target images with minimal artifacts. 3) Shape-preserving. The controllability of our method enables us to use 3D-aware landmarks as the condition during generation to preserve the shape of the source face. Extensive experiments on both FF++ and FFHQ demonstrate that our method can achieve state-of-the-art face swapping results both qualitatively and quantitatively.

With an enormous number of hand images generated over time, unleashing pose knowledge from unlabeled images for supervised hand mesh estimation is an emerging yet challenging topic. To alleviate this issue, semi-supervised and self-supervised approaches have been proposed, but they are limited by the reliance on detection models or conventional ResNet backbones. In this paper, inspired by the rapid progress of Masked Image Modeling (MIM) in visual classification tasks, we propose a novel self-supervised pre-training strategy for regressing 3D hand mesh parameters. Our approach involves a unified and multi-granularity strategy that includes a pseudo keypoint alignment module in the teacher-student framework for learning pose-aware semantic class tokens. For patch tokens with detailed locality, we adopt a self-distillation manner between teacher and student network based on MIM pre-training. To better fit low-level regression tasks, we incorporate pixel reconstruction tasks for multi-level representation learning. Additionally, we design a strong pose estimation baseline using a simple vanilla vision Transformer (ViT) as the backbone and attach a PyMAF head after tokens for regression. Extensive experiments demonstrate that our proposed approach, named HandMIM, achieves strong performance on various hand mesh estimation tasks. Notably, HandMIM outperforms specially optimized architectures, achieving 6.29mm and 8.00mm PAVPE (Vertex-Point-Error) on challenging FreiHAND and HO3Dv2 test sets, respectively, establishing new state-of-the-art records on 3D hand mesh estimation.

In this paper, we present a new approach for model acceleration by exploiting spatial sparsity in visual data. We observe that the final prediction in vision Transformers is only based on a subset of the most informative tokens, which is sufficient for accurate image recognition. Based on this observation, we propose a dynamic token sparsification framework to prune redundant tokens progressively and dynamically based on the input to accelerate vision Transformers. Specifically, we devise a lightweight prediction module to estimate the importance score of each token given the current features. The module is added to different layers to prune redundant tokens hierarchically. While the framework is inspired by our observation of the sparse attention in vision Transformers, we find the idea of adaptive and asymmetric computation can be a general solution for accelerating various architectures. We extend our method to hierarchical models including CNNs and hierarchical vision Transformers as well as more complex dense prediction tasks that require structured feature maps by formulating a more generic dynamic spatial sparsification framework with progressive sparsification and asymmetric computation for different spatial locations. By applying lightweight fast paths to less informative features and using more expressive slow paths to more important locations, we can maintain the structure of feature maps while significantly reducing the overall computations. Extensive experiments demonstrate the effectiveness of our framework on various modern architectures and different visual recognition tasks. Our results clearly demonstrate that dynamic spatial sparsification offers a new and more effective dimension for model acceleration. Code is available at https://github.com/raoyongming/DynamicViT