Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Guangtao ZhaiSuggested from co-authorship
- Patrick Le CalletSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works134 from public data
- Perceptual video quality assessment: a survey211
This survey provides a systematic overview of both classical works and recent progress in the realm of video quality assessment, which can help other researchers quickly access the field and conduct relevant research.
- 208
- Perceptual Quality Assessment of Omnidirectional Images171
A perceptual omnidirectional image quality assessment (IQA) study since it is extremely important to provide a good quality of experience under the VR environment, and some new observations different from traditional IQA are made.
- A dataset of eye movements for the children with autism spectrum disorder120
Based on this dataset, researchers could analyze the visual traits of children with ASD and design specialized visual attention models to promote research in related fields, as well as design specialized models to identify the individuals with ASD.
- 105
- Combination of near-infrared and thermal imaging techniques for the remote and simultaneous measurements of breathing and heart rates under sleep situation104
The overall performance of the proposed technique is acceptable for BR and HR estimations during nighttime and almost all the data points located within their own 95% limits of agreement are demonstrated.
- End-to-End Human-Gaze-Target Detection with Transformers96
Unlike existing two-stage methods that require human head locations as input and can predict only one human's gaze target at a time, HGTTR can directly predict the locations of all people and their gaze targets at one time in an end-to-end manner.
- Confusing Image Quality Assessment: Toward Better Augmented Reality Experience84
This paper considers AR technology as the superimposition of virtual scenes and real scenes, and introduces visual confusion as its basic theory, and establishes an ARIQA database to better simulate the real AR application scenarios.
- 73
- Large multimodal models evaluation: a survey57
By unifying general-specialized understanding and modality-specific generation evaluations, this survey clarifies the current landscape and provides guidance for future research in the LMM evaluation field.
- Visual Attention Analysis and Prediction on Human Faces for Children with Autism Spectrum Disorder55
The proposed model shows the best performance among all competitors and could be used to design specialized education contents containing human faces for the children with ASD or produce the specific model for rapidly screening ASD using their eye movement data.
- 55
- FineVQ: Fine-Grained User Generated Content Video Quality Assessment53
The first large-scale Fine-grained Video quality assessment Database is established, termed FineVD, which comprises 6104 UGC videos with fine-grained quality scores and descriptions across multiple dimensions, and a Fine-grained Video Quality assessment (FineVQ) model is proposed to learn the fine-grained quality of UGC videos.
- 51
- AIGV-Assessor: Benchmarking and Evaluating the Perceptual Quality of Text-to-Video Generation with LMM46
A novel VQA model that leverages spatiotemporal features and LMM frameworks to capture the intricate quality attributes of AIGVs, thereby accurately predicting precise video quality scores and video pair preferences is introduced and demonstrates state-of-the-art performance.
- 46
- Quality Assessment for AI Generated Images With Instruction Tuning43
The proposed MINT-IQA model achieves state-of-the-art performance in understanding and evaluating human visual preferences for AIGIs, and the proposed model also achieves competing results on traditional IQA tasks compared with state-of-the-art IQA models.
- Saliency in Augmented Reality43
To better predict the saliency in AR, a vector quantized saliency prediction method is proposed and generalize it for ARSaliency prediction, and experimental results demonstrate the superiority of the proposed method on both of the common saliency predictions problem and the AR saliency Prediction problem over benchmark methods.
- LMM4LMM: Benchmarking and Evaluating Large-Multimodal Image Generation With LMMs42
Extensive experimental results show that LMM4LMM achieves state-of-the-art performance on EvalMi-50K, and exhibits strong generalization ability on other AI-generated image evaluation benchmark datasets, manifesting the generality of both the EvalMi-50K dataset and LMM4LMM metric.
- 39
- 38
- MMHead: Towards Fine-grained Multi-modal 3D Facial Animation34
A simple but efficient VQ-VAE-based method named MM2Face is proposed to unify the multi-modal information and generate diverse and plausible 3D facial motions, which achieves competitive results on both benchmarks.
- 33
- LMM4Edit: Benchmarking and Evaluating Multimodal Image Editing with LMMs27
This work introduces EBench-18K, the first large-scale image Editing Benchmark including 18K edited images with fine-grained human preference annotations for evaluating TIE, and proposes LMM4Edit, a LMM-based metric for evaluating image Editing models from perceptual quality, editing alignment, attribute preservation, and task-specific QA accuracy in an all-in-one manner.
- 27
- 25
- 25
- MoA-VR: A Mixture-of-Agents System Toward All-in-One Video Restoration25
Extensive experiments demonstrate that MoA-VR effectively handles diverse and compound degradations, consistently outperforming existing baselines in terms of both objective metrics and perceptual quality, and highlight the potential of integrating multimodal intelligence and modular reasoning in general-purpose video restoration systems.
- 25
- 23
- AGHI-QA: A Subjective-Aligned Dataset and Metric for AI-Generated Human Images22
AGHI-QA is introduced, a large-scale benchmark specifically designed for quality assessment of AGHIs and AGHI-Assessor is proposed, 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.
- Assessment of Visually Induced Motion Sickness in Immersive Videos22
It is found that the velocity (frequency) of VOs, the time the authors immersed and the scenes’ contents will all influence the level of VIMS, and the QoE of immersive videos with real life content will be influenced correspondingly.
- FVQ: A Large-Scale Dataset and an LMM-based Method for Face Video Quality Assessment19
The first large-scale in-the-wild FVQA dataset, FVQ-20K, is presented, which contains 20,000 in-the-wild face videos together with corresponding mean opinion score (MOS) annotations and a specialized FVQA method named FVQ-Rater to achieve human-like rating and scoring for face video is proposed.
- 19
- 17
- 17
- LMME3DHF: Benchmarking and Evaluating Multimodal 3D Human Face Generation with LMMs16
Experimental results show that LMME3DHF achieves state-of-the-art performance, surpassing existing methods in both accurately predicting quality scores for AI-generated 3D human faces and effectively identifying distortion-aware salient regions and distortion types, while maintaining strong alignment with human perceptual judgments.
- HarmonyIQA: Pioneering Benchmark and Model for Image Harmonization Quality Assessment15
The first Image Quality Assessment Database for image Harmony evaluation (HarmonyIQAD) is introduced, which consists of 1,350 harmonized images generated by 9 different IHAs, and the corresponding human visual preference scores, and a Harmony Image Quality Assessment (HarmonyIQA), to predict human visual preference for harmonized images.
- LOVE: Benchmarking and Evaluating Text-to-Video Generation and Video-to-Text Interpretation15
Love is proposed, a LMM-based metric for AIGV Evaluation from multiple dimensions including perceptual preference, text-video correspondence, and task-specific accuracy in terms of both instance level and model level, which achieves state-of-the-art performance on the AIGVE-60K dataset.
- Audio-Visual Saliency for Omnidirectional Videos15
This paper establishes the largest audio-visual saliency dataset for omnidirectional videos (AVS-ODV), which comprises the omniddirectional videos, audios, and corresponding captured eye-tracking data for three video sound modalities including mute, mono, and ambisonics, and compares the performance of several state-of-the-art saliency prediction models on the dataset.
- 15
- DFBench: Benchmarking Deepfake Image Detection Capability of Large Multimodal Models14
This work proposes MoA-DF, Mixture of Agents for DeepFake detection, leveraging a combined probability strategy from multiple LMMs, which achieves state-of-the-art performance, further proving the effectiveness of leveraging LMMs for deepfake detection.
- HVEval: Towards Unified Evaluation of Human-Centric Video Generation and Understanding14
Comprehensive evaluations of 24 T2V models, 20 LMMs, and 18 VQA metrics reveal their limitations in fine-grained text-controlled generation and human-aligned perception and understanding, highlighting the significant potential of the dataset and benchmarks to advance research in human-centric video generation and understanding.
- TDVE-Assessor: Benchmarking and Evaluating the Quality of Text-Driven Video Editing with LMMs14
TDVE-Assessor is proposed, a novel VQA model specifically designed for text-driven video editing assessment that substantially outperforms existing VQA models on TDVE-DB across all three evaluation dimensions, setting a new state-of-the-art.
- ESIQA: Perceptual Quality Assessment of Vision-Pro-based Egocentric Spatial Images13
The Egocentric Spatial Images Quality Assessment Database (ESIQAD) is established, the first IQA database dedicated for egocentric spatial images as far as the authors know, and a novel mamba2-based multi-stage feature fusion model, termed ESIQAnet, which predicts the perceptual quality of egocentric spatial images under the three display modes.
- F-Bench: Rethinking Human Preference Evaluation Metrics for Benchmarking Face Generation, Customization, and Restoration13
FaceQ is introduced, the first comprehensive AI-generated Face image database with fine-grained uality annotations aligned with human preferences, and F-Bench is established, a benchmark for comparing and evaluating face generation, customization, and restoration models, highlighting strengths and weaknesses across various prompts and evaluation dimensions.
- How Does Audio Influence Visual Attention in Omnidirectional Videos? Database and Model13
An in-depth analysis of how audio influences visual attention in ODVs is performed and an innovative omnidirectional audio-visual saliency prediction network (OmniAVS) is proposed, which is built based on the U-Net architecture, and hierarchically fuses audio and visual features from the multimodal aligned embedding space.
- Masked Autoencoders as Image Processors13
This paper presents an efficient Transformer model considering both channel attention and shifted-window-based self-attention termed CSformer, and develops an effective MAE architecture for image processing (MAEIP) tasks.
- Multi-Dimensional Quality Assessment for Text-to-3D Assets: Dataset and Model12
This is the first work that studies the problem of text-guided 3D generation quality assessment, and an effective T23DAQA model is devised to evaluate the quality of AI generated 3D asset, which is more consistent with human perception.
- 11
- Binocular head-mounted chromatic pupillometry can detect structural and functional loss in glaucoma11
It is demonstrated that binocular chromatic pupillometry could potentially serve as an objective clinical tool for opportunistic glaucoma diagnosis in community-based or telemedicine-based services.
- Evaluation of the Glaucomatous Macular Damage by Chromatic Pupillometry11
The use of chromatic pupillometry offers a patient-friendly and objective approach to detect POAG, while the impairment of PLR features may serve as a potential indicator of structural macular damage.
- Omni 2 : Unifying Omnidirectional Image Generation and Editing in an Omni Model10
Any2Omni, the first comprehensive ODI generation-editing dataset, is constructed and an Omni model for Omni-directional image generation and editing ( Omni 2) is proposed, with the capability of handling various ODI generation and editing tasks under diverse input conditions using one model.
- 10
- 10
- 10
- 10
- Agentic Retoucher for Text-To-Image Generation9
Agentic Retoucher is proposed, a hierarchical decision-driven framework that reformulates post-generation correction as a human-like perception-reasoning-action loop that consistently outperforms state-of-the-art methods in perceptual quality, distortion localization and human preference alignment.
- I2I-Bench: A Comprehensive Benchmark Suite for Image-to-Image Editing Models9
This work proposes I2I-Bench, a comprehensive benchmark for image-to-image editing models, which features diverse tasks, comprehensive evaluation dimensions, and rigorous alignment validation, justifying the consistency between benchmark evaluations and human preferences.
- How is Visual Attention Influenced by Text Guidance? Database and Model9
A text-guided saliency (TGSal) prediction model, which extracts and integrates both image features and text features to predict the image saliency under various text-description conditions, which significantly outperforms the state-of-the-art saliency models on both the SJTU-TIS database and the pure image saliency databases in terms of various evaluation metrics.
- Quality-Guided Skin Tone Enhancement for Portrait Photography9
A quality-guided image enhancement paradigm is proposed that enables image enhancement models to learn the distribution of images with various quality ratings, which can be used to adjust images continuously according to different quality scores.
- 9
- GeoX-Bench: Benchmarking Cross-View Geo-Localization and Pose Estimation Capabilities of Large Multimodal Models8
The benchmark demonstrate that while current LMMs achieve impressive performance in geo-localization tasks, their effectiveness declines significantly on the more complex pose estimation tasks, highlighting a critical area for future improvement, and instruction-tuning LMMs on the training data of GeoX-Bench can significantly improve the cross-view geo-sense abilities.
- 8
- 8
- 8
- Boosting power line inspection in bad weather: Removing weather noise with channel-spatial attention-based UNet7
This paper constructs a power line components detection (PLCD) dataset, which includes 1943 power line images captured by UAVs and corresponding annotated bounding boxes, and proposes a new image restoration model termed CSUNet for better remove the noise and restore images.
- 7
- RGC-VQA: An Exploration Database for Robotic-Generated Video Quality Assessment6
The concept of Robotic-Generated Content (RGC) is innovatively proposed to term these videos generated from egocentric perspective of robots to 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.
- ESVQA: Perceptual Quality Assessment of Egocentric Spatial Videos6
The first Egocentric Spatial Video Quality Assessment Database (ESVQAD) is introduced, and a novel multi-dimensional binocular feature fusion model, termed ESVQAnet, is proposed, which integrates binocular spatial, motion, and semantic features to predict the overall perceptual quality.
- Situation-adaptive neural network for fast pre-computing image enhancement6
SAEnhancer is proposed, a situation-adaptive neural network for fast pre-computing image enhancement that learns from a small sample set to achieve personalized and adaptive enhancements for various situations without re-training, surpassing traditional 3D lookup tables (LUTs), and enhancing computational efficiency in intelligent vision applications.
- Ges-QA: A Multidimensional Quality Assessment Dataset for Audio-to-3D Gesture Generation6
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 is proposed and yields state-of-the-art performance on the Ges-QA dataset.
- 6
- Quality Assessment and Distortion-Aware Saliency Prediction for AI-Generated Omnidirectional Images5
This work first studies the quality assessment and distortion-aware saliency prediction problems for AIGODIs, and presents a corresponding optimization process that utilizes the predicted visual experience scores and distortion regions to further enhance the visual quality of an AI-generated omnidirectional image.
- ManipShield: A Unified Framework for Image Manipulation Detection, Localization and Explanation5
An all-in-one model based on a Multimodal Large Language Model (MLLM) that leverages contrastive LoRA fine-tuning and task-specific decoders to achieve unified image manipulation detection, localization, and explanation is proposed.
- LM Fight Arena: Benchmarking Large Multimodal Models via Game Competition5
LM Fight Arena (Large Model Fight Arena), a novel framework that evaluates LMMs by pitting them against each other in the classic fighting game Mortal Kombat II, is introduced, a challenging and engaging benchmark that bridges the gap between AI evaluation and interactive entertainment.
- BMPCQA: Bioinspired Metaverse Point Cloud Quality Assessment Based on Large Multimodal Models5
A bioinspired metaverse point cloud quality assessment (BMPCQA) metric, which simulates the human visual evaluation process to perform the point cloud quality assessment (PCQA) task, paves the way for perceptually aligned and scalable point cloud quality assessment in next‐generation virtual environments.
- 5
- Q-Bench-Portrait: Benchmarking Multimodal Large Language Models on Portrait Image Quality Perception4
Q-Bench-Portrait is introduced, the first holistic benchmark specifically designed for portrait image quality perception, comprising 2,765 image-question-answer triplets and 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.
- Multi-Dimensional Quality Assessment for AI-Generated Human-Centric Videos: Dataset and Model4
This work extends the previous dataset HVEval with pairwise preference annotations and proposes MoE-Rater, a Mixture-of-Experts (MoE)-inspired and multimodal large language model (MLLM)-based all-in-one method that supports multi-dimensional quality rating, multi-dimensional pairwise comparison, and category-specific question answering within a single model.
- DHQA-4D: Perceptual Quality Assessment of Dynamic 4D Digital Human4
A novel large multimodal model (LMM) based approach that is able to assess both textured 4D meshes and non-textured 4D meshes, DynaMesh-Rater, which 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.
- TIT-Score: Evaluating Long-Prompt Based Text-to-Image Alignment via Text-to-Image-to-Text Consistency4
A novel zero-shot metric based on text-to-image-to-text consistency, termed TIT, is introduced for evaluating long-prompt-generated images and achieves superior alignment with human judgment compared to CLIP-score, LMM-score, etc., with TIT-Score-LLM attaining a 7.31% absolute improvement in pairwise accuracy over the strongest baseline.
- MVBind: Self-Supervised Music Recommendation for Videos via Embedding Space Binding4
MVBind is introduced, an innovative Music-Video embedding space Binding model for cross-modal retrieval that operates as a self-supervised approach, acquiring inherent knowledge of intermodal relationships directly from data, without the need of manual annotations.
- A Saliency Dataset of Head and Eye Movements for Augmented Reality.4
To bridge the gap between the real-world scene and the scene augmented by virtual information, the ARVR saliency dataset is constructed with 100 diverse videos evaluated by 20 people and annotations of object recognition and tracking as augmented contents are blended into the omnidirectional videos.
- 3
- Evaluating image editing with LLMs: A comprehensive benchmark and intermediate-layer probing approach3
Experimental results demonstrate that widely used automatic evaluation metrics show limited correlation with human judgments on editing tasks, while EditProbe achieves substantially stronger alignment with human perception.
- 3
- 3
- 3
- 3
- 2
- LL-Bench: Rethinking Low-Level Vision Evaluation in the Era of Large-Scale Generative Models2
Extensive experiments demonstrate that LL-score not only outperforms existing image quality assessment metrics, but also serves as a promising reward model for training generative models on low-level vision tasks.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- ReasonEdit: Towards Interpretable Image Editing Evaluation via Reinforcement Learning1
ReasonEdit is developed, which is trained using reward signals derived from RE-Reward and the Group Relative Policy Optimization (GRPO) algorithm to learn an interpretable evaluation model, and achieves superior alignment with human preferences and exhibits strong generalization across public benchmarks.
- 1
- Enhancing Image Quality Assessment Ability of LMMs via Retrieval-Augmented Generation1
Extensive experiments demonstrate that the proposed IQARAG effectively boosts the IQA performance of LMMs, offering a resource-efficient alternative to fine-tuning for quality assessment.
- 1
- 1
- 1
- FMReward: Aligning and Evaluating Audio-Driven 3D Facial Animation with Human Preferences1
A comprehensive framework that learns an automatic perceptual model from human preference data and leverages it to improve and evaluate the perceptual quality of audio-driven 3D facial animation and introduces Facial Motion reward Feedback Learning (FMFL), a direct fine-tuning algorithm that leverages a pretrained reward model to optimize diffusion-based audio-driven 3D facial animation models for better alignment with human preferences.
- 1
- Facial Attractiveness Prediction in Live Streaming: A New Benchmark and Multi-modal Method1
LiveBeauty is presented, the first large-scale live-specific FAP dataset, in a more challenging application scenario, i.e., live streaming, and a multi-modal FAP method is proposed to measure the facial attractiveness in live streaming.
- 1
- 1
- Video inspired lesion segmentation algorithm for OCT retinal images1
An innovative 2.5D segmentation algorithm is proposed that treats OCT retinal images as video sequences and leverage advanced video segmentation models and significantly improves both visual quality and quantitative metrics, particularly in terms of structural continuity and robustness of the segmentation results.
- 1
- –
- Preference-guided debiasing for no-reference enhancement image quality assessment–
A preference-guided debiasing framework for no-reference enhancement image quality assessment (EIQA) that effectively mitigates algorithm-induced representation bias and achieves superior robustness and cross-algorithm generalization compared with existing approaches.
- –
- –
- –
- –
- GeoR-Bench: Evaluating Geoscience Visual Reasoning–
GeoR-Bench is presented, a Benchmark for evaluating geoscience visual visual reasoning through reasoning informed visual editing tasks, and its findings indicate that current models generate superficially plausible results but fail to capture underlying earth science processes.
- –
- –
- –
- ITIScore: An Image-to-Text-to-Image Rating Framework for the Image Captioning Ability of MLLMs–
A new large-scale image captioning benchmark is presented, termed, ICBench, which consists of both short and long captions generated by 10 advanced MLLMs on 2K images, resulting in 40K captions in total, and an automated evaluation metric is proposed, based on an image-to-text-to-image framework, which measures caption quality through reconstruction consistency.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.