Yunhao Li

Xiaoyu Huang, Zhongyu Li, Yanzhen Xiang et al. · berkeley

We present a reinforcement learning (RL) framework that enables quadrupedal robots to perform soccer goalkeeping tasks in the real world. Soccer goalkeeping using quadrupeds is a challenging problem, that combines highly dynamic locomotion with precise and fast non-prehensile object (ball) manipulation. The robot needs to react to and intercept a potentially flying ball using dynamic locomotion maneuvers in a very short amount of time, usually less than one second. In this paper, we propose to address this problem using a hierarchical model-free RL framework. The first component of the framework contains multiple control policies for distinct locomotion skills, which can be used to cover different regions of the goal. Each control policy enables the robot to track random parametric end-effector trajectories while performing one specific locomotion skill, such as jump, dive, and sidestep. These skills are then utilized by the second part of the framework which is a high-level planner to determine a desired skill and end-effector trajectory in order to intercept a ball flying to different regions of the goal. We deploy the proposed framework on a Mini Cheetah quadrupedal robot and demonstrate the effectiveness of our framework for various agile interceptions of a fast-moving ball in the real world.

Qihang Ge, Wei Sun, Yu Zhang et al.

The explosive growth of videos on streaming media platforms has underscored the urgent need for effective video quality assessment (VQA) algorithms to monitor and perceptually optimize the quality of streaming videos. However, VQA remains an extremely challenging task due to the diverse video content and the complex spatial and temporal distortions, thus necessitating more advanced methods to address these issues. Nowadays, large multimodal models (LMMs), such as GPT-4V, have exhibited strong capabilities for various visual understanding tasks, motivating us to leverage the powerful multimodal representation ability of LMMs to solve the VQA task. Therefore, we propose the first Large Multi-Modal Video Quality Assessment (LMM-VQA) model, which introduces a novel spatiotemporal visual modeling strategy for quality-aware feature extraction. Specifically, we first reformulate the quality regression problem into a question and answering (Q&A) task and construct Q&A prompts for VQA instruction tuning. Then, we design a spatiotemporal vision encoder to extract spatial and temporal features to represent the quality characteristics of videos, which are subsequently mapped into the language space by the spatiotemporal projector for modality alignment. Finally, the aligned visual tokens and the quality-inquired text tokens are aggregated as inputs for the large language model (LLM) to generate the quality score and level. Extensive experiments demonstrate that LMM-VQA achieves state-of-the-art performance across five VQA benchmarks, exhibiting an average improvement of $5\%$ in generalization ability over existing methods. Furthermore, due to the advanced design of the spatiotemporal encoder and projector, LMM-VQA also performs exceptionally well on general video understanding tasks, further validating its effectiveness. Our code will be released at https://github.com/Sueqk/LMM-VQA.

Yunhao Li, Zhenbo Yu, Yucheng Zhu et al.

Human motion synthesis is a long-standing problem with various applications in digital twins and the Metaverse. However, modern deep learning based motion synthesis approaches barely consider the physical plausibility of synthesized motions and consequently they usually produce unrealistic human motions. In order to solve this problem, we propose a system ``Skeleton2Humanoid'' which performs physics-oriented motion correction at test time by regularizing synthesized skeleton motions in a physics simulator. Concretely, our system consists of three sequential stages: (I) test time motion synthesis network adaptation, (II) skeleton to humanoid matching and (III) motion imitation based on reinforcement learning (RL). Stage I introduces a test time adaptation strategy, which improves the physical plausibility of synthesized human skeleton motions by optimizing skeleton joint locations. Stage II performs an analytical inverse kinematics strategy, which converts the optimized human skeleton motions to humanoid robot motions in a physics simulator, then the converted humanoid robot motions can be served as reference motions for the RL policy to imitate. Stage III introduces a curriculum residual force control policy, which drives the humanoid robot to mimic complex converted reference motions in accordance with the physical law. We verify our system on a typical human motion synthesis task, motion-in-betweening. Experiments on the challenging LaFAN1 dataset show our system can outperform prior methods significantly in terms of both physical plausibility and accuracy. Code will be released for research purposes at: https://github.com/michaelliyunhao/Skeleton2Humanoid

Yiwei Yang, Zheyuan Liu, Jun Jia et al.

Traditional image steganography focuses on concealing one image within another, aiming to avoid steganalysis by unauthorized entities. Coverless image steganography (CIS) enhances imperceptibility by not using any cover image. Recent works have utilized text prompts as keys in CIS through diffusion models. However, this approach faces three challenges: invalidated when private prompt is guessed, crafting public prompts for semantic diversity, and the risk of prompt leakage during frequent transmission. To address these issues, we propose DiffStega, an innovative training-free diffusion-based CIS strategy for universal application. DiffStega uses a password-dependent reference image as an image prompt alongside the text, ensuring that only authorized parties can retrieve the hidden information. Furthermore, we develop Noise Flip technique to further secure the steganography against unauthorized decryption. To comprehensively assess our method across general CIS tasks, we create a dataset comprising various image steganography instances. Experiments indicate substantial improvements in our method over existing ones, particularly in aspects of versatility, password sensitivity, and recovery quality. Codes are available at \url{https://github.com/evtricks/DiffStega}.

Yunhao Li, Sijing Wu, Zhilin Gao et al.

Large multimodal models (LMMs) have demonstrated outstanding capabilities in various visual perception tasks, which has in turn made the evaluation of LMMs significant. However, the capability of video aesthetic quality assessment, which is a fundamental ability for human, remains underexplored for LMMs. To address this, we introduce VideoAesBench, a comprehensive benchmark for evaluating LMMs' understanding of video aesthetic quality. VideoAesBench has several significant characteristics: (1) Diverse content including 1,804 videos from multiple video sources including user-generated (UGC), AI-generated (AIGC), compressed, robotic-generated (RGC), and game videos. (2) Multiple question formats containing traditional single-choice questions, multi-choice questions, True or False questions, and a novel open-ended questions for video aesthetics description. (3) Holistic video aesthetics dimensions including visual form related questions from 5 aspects, visual style related questions from 4 aspects, and visual affectiveness questions from 3 aspects. Based on VideoAesBench, we benchmark 23 open-source and commercial large multimodal models. Our findings show that current LMMs only contain basic video aesthetics perception ability, their performance remains incomplete and imprecise. We hope our VideoAesBench can be served as a strong testbed and offer insights for explainable video aesthetics assessment. The data will be released on https://github.com/michaelliyunhao/VideoAesBench

Sijing Wu, Yunhao Li, Zicheng Zhang et al.

Recent advances in multimodal large language models (MLLMs) have demonstrated impressive performance on existing low-level vision benchmarks, which primarily focus on generic images. However, their capabilities to perceive and assess portrait images, a domain characterized by distinct structural and perceptual properties, remain largely underexplored. To this end, we introduce Q-Bench-Portrait, the first holistic benchmark specifically designed for portrait image quality perception, comprising 2,765 image-question-answer triplets and featuring (1) diverse portrait image sources, including natural, synthetic distortion, AI-generated, artistic, and computer graphics images; (2) comprehensive quality dimensions, covering technical distortions, AIGC-specific distortions, and aesthetics; and (3) a range of question formats, including single-choice, multiple-choice, true/false, and open-ended questions, at both global and local levels. Based on Q-Bench-Portrait, we evaluate 20 open-source and 5 closed-source MLLMs, revealing that although current models demonstrate some competence in portrait image perception, their performance remains limited and imprecise, with a clear gap relative to human judgments. We hope that the proposed benchmark will foster further research into enhancing the portrait image perception capabilities of both general-purpose and domain-specific MLLMs.

Sijing Wu, Yichao Yan, Yunhao Li et al.

To bring digital avatars into people's lives, it is highly demanded to efficiently generate complete, realistic, and animatable head avatars. This task is challenging, and it is difficult for existing methods to satisfy all the requirements at once. To achieve these goals, we propose GANHead (Generative Animatable Neural Head Avatar), a novel generative head model that takes advantages of both the fine-grained control over the explicit expression parameters and the realistic rendering results of implicit representations. Specifically, GANHead represents coarse geometry, fine-gained details and texture via three networks in canonical space to obtain the ability to generate complete and realistic head avatars. To achieve flexible animation, we define the deformation filed by standard linear blend skinning (LBS), with the learned continuous pose and expression bases and LBS weights. This allows the avatars to be directly animated by FLAME parameters and generalize well to unseen poses and expressions. Compared to state-of-the-art (SOTA) methods, GANHead achieves superior performance on head avatar generation and raw scan fitting.

Yunhao Li, Zhen Xiao, Lin Yang et al.

Multi-object tracking (MOT) is a fundamental problem in computer vision with numerous applications, such as intelligent surveillance and automated driving. Despite the significant progress made in MOT, pedestrian attributes, such as gender, hairstyle, body shape, and clothing features, which contain rich and high-level information, have been less explored. To address this gap, we propose a simple, effective, and generic method to predict pedestrian attributes to support general Re-ID embedding. We first introduce AttMOT, a large, highly enriched synthetic dataset for pedestrian tracking, containing over 80k frames and 6 million pedestrian IDs with different time, weather conditions, and scenarios. To the best of our knowledge, AttMOT is the first MOT dataset with semantic attributes. Subsequently, we explore different approaches to fuse Re-ID embedding and pedestrian attributes, including attention mechanisms, which we hope will stimulate the development of attribute-assisted MOT. The proposed method AAM demonstrates its effectiveness and generality on several representative pedestrian multi-object tracking benchmarks, including MOT17 and MOT20, through experiments on the AttMOT dataset. When applied to state-of-the-art trackers, AAM achieves consistent improvements in MOTA, HOTA, AssA, IDs, and IDF1 scores. For instance, on MOT17, the proposed method yields a +1.1 MOTA, +1.7 HOTA, and +1.8 IDF1 improvement when used with FairMOT. To encourage further research on attribute-assisted MOT, we will release the AttMOT dataset.

Wenjin Wang, Yunhao Li, Yixin Ou et al.

Layout-aware pre-trained models has achieved significant progress on document image question answering. They introduce extra learnable modules into existing language models to capture layout information within document images from text bounding box coordinates obtained by OCR tools. However, extra modules necessitate pre-training on extensive document images. This prevents these methods from directly utilizing off-the-shelf instruction-tuning language foundation models, which have recently shown promising potential in zero-shot learning. Instead, in this paper, we find that instruction-tuning language models like Claude and ChatGPT can understand layout by spaces and line breaks. Based on this observation, we propose the LAyout and Task aware Instruction Prompt (LATIN-Prompt), which consists of layout-aware document content and task-aware instruction. Specifically, the former uses appropriate spaces and line breaks to recover the layout information among text segments obtained by OCR tools, and the latter ensures that generated answers adhere to formatting requirements. Moreover, we propose the LAyout and Task aware Instruction Tuning (LATIN-Tuning) to improve the performance of small instruction-tuning models like Alpaca. Experimental results show that LATIN-Prompt enables zero-shot performance of Claude and ChatGPT to be comparable to the fine-tuning performance of SOTAs on document image question answering, and LATIN-Tuning enhances the zero-shot performance of Alpaca significantly. For example, LATIN-Prompt improves the performance of Claude and ChatGPT on DocVQA by 263% and 20% respectively. LATIN-Tuning improves the performance of Alpaca on DocVQA by 87.7%. Quantitative and qualitative analyses demonstrate the effectiveness of LATIN-Prompt and LATIN-Tuning. We provide the code in supplementary and will release it to facilitate future research.

Nan Jiang, Zimo He, Wanhe Yu et al. · pku

A long-standing objective in humanoid robotics is the realization of versatile agents capable of following diverse multimodal instructions with human-level flexibility. Despite advances in humanoid control, bridging high-level multimodal perception with whole-body execution remains a significant bottleneck. Existing methods often struggle to translate heterogeneous instructions -- such as language, music, and trajectories -- into stable, real-time actions. Here we show that UniAct, a two-stage framework integrating a fine-tuned MLLM with a causal streaming pipeline, enables humanoid robots to execute multimodal instructions with sub-500 ms latency. By unifying inputs through a shared discrete codebook via FSQ, UniAct ensures cross-modal alignment while constraining motions to a physically grounded manifold. This approach yields a 19% improvement in the success rate of zero-shot tracking of imperfect reference motions. We validate UniAct on UniMoCap, our 20-hour humanoid motion benchmark, demonstrating robust generalization across diverse real-world scenarios. Our results mark a critical step toward responsive, general-purpose humanoid assistants capable of seamless interaction through unified perception and control.

Zhilin Gao, Yunhao Li, Sijing Wu et al.

The Audio-to-3D-Gesture (A2G) task has enormous potential for various applications in virtual reality and computer graphics, etc. However, current evaluation metrics, such as Fréchet Gesture Distance or Beat Constancy, fail at reflecting the human preference of the generated 3D gestures. To cope with this problem, exploring human preference and an objective quality assessment metric for AI-generated 3D human gestures is becoming increasingly significant. In this paper, we introduce the Ges-QA dataset, which includes 1,400 samples with multidimensional scores for gesture quality and audio-gesture consistency. Moreover, we collect binary classification labels to determine whether the generated gestures match the emotions of the audio. Equipped with our Ges-QA dataset, we propose a multi-modal transformer-based neural network with 3 branches for video, audio and 3D skeleton modalities, which can score A2G contents in multiple dimensions. Comparative experimental results and ablation studies demonstrate that Ges-QAer yields state-of-the-art performance on our dataset.

Yunhao Li, Qin Li, Hao Wang et al.

Current multi-object tracking (MOT) aims to predict trajectories of targets (i.e., ''where'') in videos. Yet, knowing merely ''where'' is insufficient in many crucial applications. In comparison, semantic understanding such as fine-grained behaviors, interactions, and overall summarized captions (i.e., ''what'') from videos, associated with ''where'', is highly-desired for comprehensive video analysis. Thus motivated, we introduce Semantic Multi-Object Tracking (SMOT), that aims to estimate object trajectories and meanwhile understand semantic details of associated trajectories including instance captions, instance interactions, and overall video captions, integrating ''where'' and ''what'' for tracking. In order to foster the exploration of SMOT, we propose BenSMOT, a large-scale Benchmark for Semantic MOT. Specifically, BenSMOT comprises 3,292 videos with 151K frames, covering various scenarios for semantic tracking of humans. BenSMOT provides annotations for the trajectories of targets, along with associated instance captions in natural language, instance interactions, and overall caption for each video sequence. To our best knowledge, BenSMOT is the first publicly available benchmark for SMOT. Besides, to encourage future research, we present a novel tracker named SMOTer, which is specially designed and end-to-end trained for SMOT, showing promising performance. By releasing BenSMOT, we expect to go beyond conventional MOT by predicting ''where'' and ''what'' for SMOT, opening up a new direction in tracking for video understanding. We will release BenSMOT and SMOTer at https://github.com/Nathan-Li123/SMOTer.

Zhichao Zhang, Wei Sun, Xinyue Li et al.

AI-driven video generation techniques have made significant progress in recent years. However, AI-generated videos (AGVs) involving human activities often exhibit substantial visual and semantic distortions, hindering the practical application of video generation technologies in real-world scenarios. To address this challenge, we conduct a pioneering study on human activity AGV quality assessment, focusing on visual quality evaluation and the identification of semantic distortions. First, we construct the AI-Generated Human activity Video Quality Assessment (Human-AGVQA) dataset, consisting of 6,000 AGVs derived from 15 popular text-to-video (T2V) models using 400 text prompts that describe diverse human activities. We conduct a subjective study to evaluate the human appearance quality, action continuity quality, and overall video quality of AGVs, and identify semantic issues of human body parts. Based on Human-AGVQA, we benchmark the performance of T2V models and analyze their strengths and weaknesses in generating different categories of human activities. Second, we develop an objective evaluation metric, named AI-Generated Human activity Video Quality metric (GHVQ), to automatically analyze the quality of human activity AGVs. GHVQ systematically extracts human-focused quality features, AI-generated content-aware quality features, and temporal continuity features, making it a comprehensive and explainable quality metric for human activity AGVs. The extensive experimental results show that GHVQ outperforms existing quality metrics on the Human-AGVQA dataset by a large margin, demonstrating its efficacy in assessing the quality of human activity AGVs. The Human-AGVQA dataset and GHVQ metric will be released at https://github.com/zczhang-sjtu/GHVQ.git.

Yunhao Li, Xiaodong Wang, Ping Wang et al.

In this paper, we explore the potential of Snapshot Compressive Imaging (SCI) technique for recovering the underlying 3D scene representation from a single temporal compressed image. SCI is a cost-effective method that enables the recording of high-dimensional data, such as hyperspectral or temporal information, into a single image using low-cost 2D imaging sensors. To achieve this, a series of specially designed 2D masks are usually employed, which not only reduces storage requirements but also offers potential privacy protection. Inspired by this, to take one step further, our approach builds upon the powerful 3D scene representation capabilities of neural radiance fields (NeRF). Specifically, we formulate the physical imaging process of SCI as part of the training of NeRF, allowing us to exploit its impressive performance in capturing complex scene structures. To assess the effectiveness of our method, we conduct extensive evaluations using both synthetic data and real data captured by our SCI system. Extensive experimental results demonstrate that our proposed approach surpasses the state-of-the-art methods in terms of image reconstruction and novel view image synthesis. Moreover, our method also exhibits the ability to restore high frame-rate multi-view consistent images by leveraging SCI and the rendering capabilities of NeRF. The code is available at https://github.com/WU-CVGL/SCINeRF.

Jianing Jin, Jiangyong Ying, Huiyu Duan et al.

As camera-equipped robotic platforms become increasingly integrated into daily life, robotic-generated videos have begun to appear on streaming media platforms, enabling us to envision a future where humans and robots coexist. We innovatively propose the concept of Robotic-Generated Content (RGC) to term these videos generated from egocentric perspective of robots. The perceptual quality of RGC videos is critical in human-robot interaction scenarios, and RGC videos exhibit unique distortions and visual requirements that differ markedly from those of professionally-generated content (PGC) videos and user-generated content (UGC) videos. However, dedicated research on quality assessment of RGC videos is still lacking. To address this gap and to support broader robotic applications, we establish the first Robotic-Generated Content Database (RGCD), which contains a total of 2,100 videos drawn from three robot categories and sourced from diverse platforms. A subjective VQA experiment is conducted subsequently to assess human visual perception of robotic-generated videos. Finally, we conduct a benchmark experiment to evaluate the performance of 11 state-of-the-art VQA models on our database. Experimental results reveal significant limitations in existing VQA models when applied to complex, robotic-generated content, highlighting a critical need for RGC-specific VQA models. Our RGCD is publicly available at: https://github.com/IntMeGroup/RGC-VQA.

Liu Yang, Huiyu Duan, Ran Tao et al.

Omnidirectional images (ODIs) provide full 360x180 view which are widely adopted in VR, AR and embodied intelligence applications. While multi-modal large language models (MLLMs) have demonstrated remarkable performance on conventional 2D image and video understanding benchmarks, their ability to comprehend the immersive environments captured by ODIs remains largely unexplored. To address this gap, we first present ODI-Bench, a novel comprehensive benchmark specifically designed for omnidirectional image understanding. ODI-Bench contains 2,000 high-quality omnidirectional images and over 4,000 manually annotated question-answering (QA) pairs across 10 fine-grained tasks, covering both general-level and spatial-level ODI understanding. Extensive experiments are conducted to benchmark 20 representative MLLMs, including proprietary and open-source models, under both close-ended and open-ended settings. Experimental results reveal that current MLLMs still struggle to capture the immersive context provided by ODIs. To this end, we further introduce Omni-CoT, a training-free method which significantly enhances MLLMs' comprehension ability in the omnidirectional environment through chain-of-thought reasoning across both textual information and visual cues. Both the benchmark and the code will be released upon the publication.

Yunhao Li, Xiaoqiong Liu, Luke Liu et al.

Vision-Language MOT is a crucial tracking problem and has drawn increasing attention recently. It aims to track objects based on human language commands, replacing the traditional use of templates or pre-set information from training sets in conventional tracking tasks. Despite various efforts, a key challenge lies in the lack of a clear understanding of why language is used for tracking, which hinders further development in this field. In this paper, we address this challenge by introducing Language-Guided MOT, a unified task framework, along with a corresponding large-scale benchmark, termed LaMOT, which encompasses diverse scenarios and language descriptions. Specially, LaMOT comprises 1,660 sequences from 4 different datasets and aims to unify various Vision-Language MOT tasks while providing a standardized evaluation platform. To ensure high-quality annotations, we manually assign appropriate descriptive texts to each target in every video and conduct careful inspection and correction. To the best of our knowledge, LaMOT is the first benchmark dedicated to Language-Guided MOT. Additionally, we propose a simple yet effective tracker, termed LaMOTer. By establishing a unified task framework, providing challenging benchmarks, and offering insights for future algorithm design and evaluation, we expect to contribute to the advancement of research in Vision-Language MOT. We will release the data at https://github.com/Nathan-Li123/LaMOT.

Yifan Jiao, Yunhao Li, Junhua Ding et al.

In this paper, we present a novel benchmark, GSOT3D, that aims at facilitating development of generic 3D single object tracking (SOT) in the wild. Specifically, GSOT3D offers 620 sequences with 123K frames, and covers a wide selection of 54 object categories. Each sequence is offered with multiple modalities, including the point cloud (PC), RGB image, and depth. This allows GSOT3D to support various 3D tracking tasks, such as single-modal 3D SOT on PC and multi-modal 3D SOT on RGB-PC or RGB-D, and thus greatly broadens research directions for 3D object tracking. To provide highquality per-frame 3D annotations, all sequences are labeled manually with multiple rounds of meticulous inspection and refinement. To our best knowledge, GSOT3D is the largest benchmark dedicated to various generic 3D object tracking tasks. To understand how existing 3D trackers perform and to provide comparisons for future research on GSOT3D, we assess eight representative point cloud-based tracking models. Our evaluation results exhibit that these models heavily degrade on GSOT3D, and more efforts are required for robust and generic 3D object tracking. Besides, to encourage future research, we present a simple yet effective generic 3D tracker, named PROT3D, that localizes the target object via a progressive spatial-temporal network and outperforms all current solutions by a large margin. By releasing GSOT3D, we expect to advance further 3D tracking in future research and applications. Our benchmark and model as well as the evaluation results will be publicly released at our webpage https://github.com/ailovejinx/GSOT3D.

Wei Luo, Yiting Lu, Xin Li et al.

This paper reports on the LoViF 2026 PhyScore challenge, a competition on holistic quality assessment of world-model-generated videos across both 2D and 4D generation settings. The challenge is motivated by a central gap in current evaluation practice: perceptual quality alone is insufficient to judge whether generated dynamics are physically plausible, temporally coherent, and consistent with input conditions. Participants are required to build a metric that jointly predicts four dimensions, i.e., Video Quality, Physical Realism, Condition-Video Alignment, and Temporal Consistency. Depart from that, participants also need to localize physical anomaly timestamps for fine-grained diagnosis. The benchmark dataset contains 1,554 videos generated by seven representative world generative models, organized into three tracks (text-2D, image-to-4D, and video-to-4D) and spanning 26 categories. These categories explicitly cover physics-relevant scenarios, including dynamics, optics, and thermodynamics, together with diverse real-world and creative content. To ensure label reliability, scores and anomaly timestamps are produced through trained human annotation with an additional automated quality-control pass. Evaluation is based on both score prediction and anomaly localization, with a composite protocol that combines TimeStamp_IOU and SRCC/PLCC. This report summarizes the challenge design and provides method-level insights from submitted solutions.

Yang Zhan, Yunhao Li, Zhang Chao et al.

Recent advancements in reinforcement fine-tuning have significantly improved the reasoning ability of large language models (LLMs). In particular, methods such as group relative policy optimization (GRPO) have demonstrated strong capabilities across various fields. However, applying LLMs to ship trajectory prediction remains largely unexplored. In this paper, we propose ShipTraj-R1, a novel LLM-based framework that reformulates ship trajectory prediction as a text-to-text generation problem. (1) We design a dynamic prompt containing trajectory information about conflicting ships to guide the model to achieve adaptive chain-of-thought (CoT) reasoning. (2) We introduce a comprehensive rule-based reward mechanism to incentivize the reasoning format and prediction accuracy of the model. (3) Our ShipTraj-R1 is reinforced through the GRPO mechanism guided by domain-specific prompts and rewards, and utilizes the Qwen3 as the model backbone. Extensive experimental results on two complex and real-world maritime datasets show that the proposed ShipTraj-R1 achieves the least error compared with state-of-the-art deep learning and LLM-based baselines.

Shuhao Han, Haotian Fan, Fangyuan Kong et al.

This paper reports on the NTIRE 2025 challenge on Text to Image (T2I) generation model quality assessment, which will be held in conjunction with the New Trends in Image Restoration and Enhancement Workshop (NTIRE) at CVPR 2025. The aim of this challenge is to address the fine-grained quality assessment of text-to-image generation models. This challenge evaluates text-to-image models from two aspects: image-text alignment and image structural distortion detection, and is divided into the alignment track and the structural track. The alignment track uses the EvalMuse-40K, which contains around 40K AI-Generated Images (AIGIs) generated by 20 popular generative models. The alignment track has a total of 371 registered participants. A total of 1,883 submissions are received in the development phase, and 507 submissions are received in the test phase. Finally, 12 participating teams submitted their models and fact sheets. The structure track uses the EvalMuse-Structure, which contains 10,000 AI-Generated Images (AIGIs) with corresponding structural distortion mask. A total of 211 participants have registered in the structure track. A total of 1155 submissions are received in the development phase, and 487 submissions are received in the test phase. Finally, 8 participating teams submitted their models and fact sheets. Almost all methods have achieved better results than baseline methods, and the winning methods in both tracks have demonstrated superior prediction performance on T2I model quality assessment.

Sijing Wu, Yunhao Li, Ziwen Xu et al.

Face video quality assessment (FVQA) deserves to be explored in addition to general video quality assessment (VQA), as face videos are the primary content on social media platforms and human visual system (HVS) is particularly sensitive to human faces. However, FVQA is rarely explored due to the lack of large-scale FVQA datasets. To fill this gap, we present the first large-scale in-the-wild FVQA dataset, FVQ-20K, which contains 20,000 in-the-wild face videos together with corresponding mean opinion score (MOS) annotations. Along with the FVQ-20K dataset, we further propose a specialized FVQA method named FVQ-Rater to achieve human-like rating and scoring for face video, which is the first attempt to explore the potential of large multimodal models (LMMs) for the FVQA task. Concretely, we elaborately extract multi-dimensional features including spatial features, temporal features, and face-specific features (i.e., portrait features and face embeddings) to provide comprehensive visual information, and take advantage of the LoRA-based instruction tuning technique to achieve quality-specific fine-tuning, which shows superior performance on both FVQ-20K and CFVQA datasets. Extensive experiments and comprehensive analysis demonstrate the significant potential of the FVQ-20K dataset and FVQ-Rater method in promoting the development of FVQA.

Yunhao Li, Sijing Wu, Wei Sun et al.

The rapid development of text-to-image (T2I) generation approaches has attracted extensive interest in evaluating the quality of generated images, leading to the development of various quality assessment methods for general-purpose T2I outputs. However, existing image quality assessment (IQA) methods are limited to providing global quality scores, failing to deliver fine-grained perceptual evaluations for structurally complex subjects like humans, which is a critical challenge considering the frequent anatomical and textural distortions in AI-generated human images (AGHIs). To address this gap, we introduce AGHI-QA, the first large-scale benchmark specifically designed for quality assessment of AGHIs. The dataset comprises 4,000 images generated from 400 carefully crafted text prompts using 10 state of-the-art T2I models. We conduct a systematic subjective study to collect multidimensional annotations, including perceptual quality scores, text-image correspondence scores, visible and distorted body part labels. Based on AGHI-QA, we evaluate the strengths and weaknesses of current T2I methods in generating human images from multiple dimensions. Furthermore, we propose AGHI-Assessor, a novel quality metric that integrates the large multimodal model (LMM) with domain-specific human features for precise quality prediction and identification of visible and distorted body parts in AGHIs. Extensive experimental results demonstrate that AGHI-Assessor showcases state-of-the-art performance, significantly outperforming existing IQA methods in multidimensional quality assessment and surpassing leading LMMs in detecting structural distortions in AGHIs.

Yunhao Li, Jing Wu, Lingzhe Zhao et al.

When capturing images through the glass during rainy or snowy weather conditions, the resulting images often contain waterdrops adhered on the glass surface, and these waterdrops significantly degrade the image quality and performance of many computer vision algorithms. To tackle these limitations, we propose a method to reconstruct the clear 3D scene implicitly from multi-view images degraded by waterdrops. Our method exploits an attention network to predict the location of waterdrops and then train a Neural Radiance Fields to recover the 3D scene implicitly. By leveraging the strong scene representation capabilities of NeRF, our method can render high-quality novel-view images with waterdrops removed. Extensive experimental results on both synthetic and real datasets show that our method is able to generate clear 3D scenes and outperforms existing state-of-the-art (SOTA) image adhesive waterdrop removal methods.

Sijing Wu, Yunhao Li, Weitian Zhang et al.

Singing, as a common facial movement second only to talking, can be regarded as a universal language across ethnicities and cultures, plays an important role in emotional communication, art, and entertainment. However, it is often overlooked in the field of audio-driven facial animation due to the lack of singing head datasets and the domain gap between singing and talking in rhythm and amplitude. To this end, we collect a high-quality large-scale singing head dataset, SingingHead, which consists of more than 27 hours of synchronized singing video, 3D facial motion, singing audio, and background music from 76 individuals and 8 types of music. Along with the SingingHead dataset, we benchmark existing audio-driven 3D facial animation methods and 2D talking head methods on the singing task. Furthermore, we argue that 3D and 2D facial animation tasks can be solved together, and propose a unified singing head animation framework named UniSinger to achieve both singing audio-driven 3D singing head animation and 2D singing portrait video synthesis, which achieves competitive results on both 3D and 2D benchmarks. Extensive experiments demonstrate the significance of the proposed singing-specific dataset in promoting the development of singing head animation tasks, as well as the promising performance of our unified facial animation framework.

Yunhao Li, Sijing Wu, Huiyu Duan et al.

In recent years, with the rapid development of powerful multimodal large language models (MLLMs), explainable image quality assessment (IQA) has gradually become popular, aiming at providing quality-related descriptions and answers of images. To achieve this goal, recent methods seek to construct a large-scale instruction tuning dataset to empower the MLLM with quality perception ability following the well-known scaling law. However, a large amount of instruction tuning data may cause substantial computational costs and redundant data, which in turn will cause harm to the performance of the model. To cope with this problem, in this paper, we challenge the scaling law and systematically investigate the role of data quality of the instruction tuning dataset for explainable IQA. Using a powerful pre-trained MLLM, we first investigate the changes in model performance after fine-tuning with different sizes of instruction tuning data. We find that selecting a subset of the data set randomly using an appropriate ratio can even lead to better results than training with the entire instruction tuning dataset, demonstrating the redundancy of current explainable IQA instruction tuning data. Beyond randomly sampling a subset, we propose a clustering-based data selection framework with three stages: clustering feature extraction, cluster quota allocation, and cluster sampling strategy. Then we systematically analyze the choices of each stage and propose a simple but efficient data selection method IQA-Select for explainable IQA. The experimental results demonstrate that IQA-Select can achieve 102.1% and 103.7% performance of full fine-tuning using only 10% selected data in Q-Bench and AesBench respectively, significantly reducing computational costs while achieving better performance.

Yunhao Li, Sijing Wu, Yucheng Zhu et al.

With the rapid development of 3D scanning and reconstruction technologies, dynamic digital human avatars based on 4D meshes have become increasingly popular. A high-precision dynamic digital human avatar can be applied to various fields such as game production, animation generation, and remote immersive communication. However, these 4D human avatar meshes are prone to being degraded by various types of noise during the processes of collection, compression, and transmission, thereby affecting the viewing experience of users. In light of this fact, quality assessment of dynamic 4D digital humans becomes increasingly important. In this paper, we first propose a large-scale dynamic digital human quality assessment dataset, DHQA-4D, which contains 32 high-quality real-scanned 4D human mesh sequences, 1920 distorted textured 4D human meshes degraded by 11 textured distortions, as well as their corresponding textured and non-textured mean opinion scores (MOSs). Equipped with DHQA-4D dataset, we analyze the influence of different types of distortion on human perception for textured dynamic 4D meshes and non-textured dynamic 4D meshes. Additionally, we propose DynaMesh-Rater, a novel large multimodal model (LMM) based approach that is able to assess both textured 4D meshes and non-textured 4D meshes. Concretely, DynaMesh-Rater elaborately extracts multi-dimensional features, including visual features from a projected 2D video, motion features from cropped video clips, and geometry features from the 4D human mesh to provide comprehensive quality-related information. Then we utilize a LMM model to integrate the multi-dimensional features and conduct a LoRA-based instruction tuning technique to teach the LMM model to predict the quality scores. Extensive experimental results on the DHQA-4D dataset demonstrate the superiority of our DynaMesh-Rater method over previous quality assessment methods.

Li Gao, Hongyang Sun, Liu Liu et al.

Accurate visual localization is crucial for autonomous driving, yet existing methods face a fundamental dilemma: While high-definition (HD) maps provide high-precision localization references, their costly construction and maintenance hinder scalability, which drives research toward standard-definition (SD) maps like OpenStreetMap. Current SD-map-based approaches primarily focus on Bird's-Eye View (BEV) matching between images and maps, overlooking a ubiquitous signal-noisy GPS. Although GPS is readily available, it suffers from multipath errors in urban environments. We propose DiffVL, the first framework to reformulate visual localization as a GPS denoising task using diffusion models. Our key insight is that noisy GPS trajectory, when conditioned on visual BEV features and SD maps, implicitly encode the true pose distribution, which can be recovered through iterative diffusion refinement. DiffVL, unlike prior BEV-matching methods (e.g., OrienterNet) or transformer-based registration approaches, learns to reverse GPS noise perturbations by jointly modeling GPS, SD map, and visual signals, achieving sub-meter accuracy without relying on HD maps. Experiments on multiple datasets demonstrate that our method achieves state-of-the-art accuracy compared to BEV-matching baselines. Crucially, our work proves that diffusion models can enable scalable localization by treating noisy GPS as a generative prior-making a paradigm shift from traditional matching-based methods.

Huadong Wu, Yi Fu, Yunhao Li et al.

Facial makeup editing aims to realistically transfer makeup from a reference to a target face. Existing methods often produce low-quality results with coarse makeup details and struggle to preserve both identity and makeup fidelity, mainly due to the lack of structured paired data -- where source and result share identity, and reference and result share identical makeup. To address this, we introduce MakeupQuad, a large-scale, high-quality dataset with non-makeup faces, references, edited results, and textual makeup descriptions. Building on this, we propose EvoMakeup, a unified training framework that mitigates image degradation during multi-stage distillation, enabling iterative improvement of both data and model quality. Although trained solely on synthetic data, EvoMakeup generalizes well and outperforms prior methods on real-world benchmarks. It supports high-fidelity, controllable, multi-task makeup editing -- including full-face and partial reference-based editing, as well as text-driven makeup editing -- within a single model. Experimental results demonstrate that our method achieves superior makeup fidelity and identity preservation, effectively balancing both aspects. Code and dataset will be released upon acceptance.

Yunhao Li, Yifan Jiao, Dan Meng et al.

Open-Vocabulary Multi-Object Tracking (OV-MOT) aims to enable approaches to track objects without being limited to a predefined set of categories. Current OV-MOT methods typically rely primarily on instance-level detection and association, often overlooking trajectory information that is unique and essential for object tracking tasks. Utilizing trajectory information can enhance association stability and classification accuracy, especially in cases of occlusion and category ambiguity, thereby improving adaptability to novel classes. Thus motivated, in this paper we propose \textbf{TRACT}, an open-vocabulary tracker that leverages trajectory information to improve both object association and classification in OV-MOT. Specifically, we introduce a \textit{Trajectory Consistency Reinforcement} (\textbf{TCR}) strategy, that benefits tracking performance by improving target identity and category consistency. In addition, we present \textbf{TraCLIP}, a plug-and-play trajectory classification module. It integrates \textit{Trajectory Feature Aggregation} (\textbf{TFA}) and \textit{Trajectory Semantic Enrichment} (\textbf{TSE}) strategies to fully leverage trajectory information from visual and language perspectives for enhancing the classification results. Extensive experiments on OV-TAO show that our TRACT significantly improves tracking performance, highlighting trajectory information as a valuable asset for OV-MOT. Code will be released.

Yunhao Li, Xiang Liu, Xiaodong Wang et al.

In this paper, we explore the potential of Snapshot Compressive Imaging (SCI) technique for recovering the underlying 3D scene structure from a single temporal compressed image. SCI is a cost-effective method that enables the recording of high-dimensional data, such as hyperspectral or temporal information, into a single image using low-cost 2D imaging sensors. To achieve this, a series of specially designed 2D masks are usually employed, reducing storage and transmission requirements and offering potential privacy protection. Inspired by this, we take one step further to recover the encoded 3D scene information leveraging powerful 3D scene representation capabilities of neural radiance fields (NeRF). Specifically, we propose SCINeRF, in which we formulate the physical imaging process of SCI as part of the training of NeRF, allowing us to exploit its impressive performance in capturing complex scene structures. In addition, we further integrate the popular 3D Gaussian Splatting (3DGS) framework and propose SCISplat to improve 3D scene reconstruction quality and training/rendering speed by explicitly optimizing point clouds into 3D Gaussian representations. To assess the effectiveness of our method, we conduct extensive evaluations using both synthetic data and real data captured by our SCI system. Experimental results demonstrate that our proposed approach surpasses the state-of-the-art methods in terms of image reconstruction and novel view synthesis. Moreover, our method also exhibits the ability to render high frame-rate multi-view consistent images in real time by leveraging SCI and the rendering capabilities of 3DGS. Codes will be available at: https://github.com/WU- CVGL/SCISplat.

Yunhao Li, Yunyi Yang, Xiaojun Quan et al.

Dialogue policy learning, a subtask that determines the content of system response generation and then the degree of task completion, is essential for task-oriented dialogue systems. However, the unbalanced distribution of system actions in dialogue datasets often causes difficulty in learning to generate desired actions and responses. In this paper, we propose a retrieve-and-memorize framework to enhance the learning of system actions. Specially, we first design a neural context-aware retrieval module to retrieve multiple candidate system actions from the training set given a dialogue context. Then, we propose a memory-augmented multi-decoder network to generate the system actions conditioned on the candidate actions, which allows the network to adaptively select key information in the candidate actions and ignore noises. We conduct experiments on the large-scale multi-domain task-oriented dialogue dataset MultiWOZ 2.0 and MultiWOZ 2.1. Experimental results show that our method achieves competitive performance among several state-of-the-art models in the context-to-response generation task.

Merlin A. Nau, Florian Schiffers, Yunhao Li et al.

The utilization of computational photography becomes increasingly essential in the medical field. Today, imaging techniques for dermatology range from two-dimensional (2D) color imagery with a mobile device to professional clinical imaging systems measuring additional detailed three-dimensional (3D) data. The latter are commonly expensive and not accessible to a broad audience. In this work, we propose a novel system and software framework that relies only on low-cost (and even mobile) commodity devices present in every household to measure detailed 3D information of the human skin with a 3D-gradient-illumination-based method. We believe that our system has great potential for early-stage diagnosis and monitoring of skin diseases, especially in vastly populated or underdeveloped areas.

Yunyi Yang, Yunhao Li, Xiaojun Quan

This paper presents our task-oriented dialog system UBAR which models task-oriented dialogs on a dialog session level. Specifically, UBAR is acquired by fine-tuning the large pre-trained unidirectional language model GPT-2 on the sequence of the entire dialog session which is composed of user utterance, belief state, database result, system act, and system response of every dialog turn. Additionally, UBAR is evaluated in a more realistic setting, where its dialog context has access to user utterances and all content it generated such as belief states, system acts, and system responses. Experimental results on the MultiWOZ datasets show that UBAR achieves state-of-the-art performances in multiple settings, improving the combined score of response generation, policy optimization, and end-to-end modeling by 4.7, 3.5, and 9.4 points respectively. Thorough analyses demonstrate that the session-level training sequence formulation and the generated dialog context are essential for UBAR to operate as a fully end-to-end task-oriented dialog system in real life. We also examine the transfer ability of UBAR to new domains with limited data and provide visualization and a case study to illustrate the advantages of UBAR in modeling on a dialog session level.