Reinforcement fine-tuning (RFT) is a proliferating paradigm for LMM training. Analogous to high-level reasoning tasks, RFT is similarly applicable to low-level vision domains, including image quality assessment (IQA). Existing RFT-based IQA methods typically use rule-based output rewards to verify the model's rollouts but provide no reward supervision for the "think” process, leaving its correctness and efficacy uncontrolled. Furthermore, these methods typically fine-tune directly on downstream IQA tasks without explicitly enhancing the model’s native low-level visual quality perception, which may constrain its performance upper bound. In response to these gaps, we propose the multi‐stage RFT IQA framework (Refine-IQA). In Stage-1, we build the Refine-Perception-20K dataset (with 12 main distortions, 20,907 locally-distorted images, and over 55K RFT samples) and design multi-task reward functions to strengthen the model’s visual quality perception. In Stage-2, targeting the quality scoring task, we introduce a probability difference reward involved strategy for "think" process supervision. The resulting Refine-IQA Series Models achieve outstanding performance on both perception and scoring tasks—and, notably, our paradigm activates a robust "think” (quality interpretating) capability that also attains exceptional results on the corresponding quality interpreting benchmark.

The data scaling law has significantly enhanced large multi-modal models (LMMs) performance across various downstream tasks. However, in the domain of perceptual video quality assessment (VQA), the potential of data scaling remains unprecedented due to the scarcity of labeled resources and the insufficient scale of datasets. To address this, we propose OmniVQA, a framework designed to efficiently build high-quality, machine-dominated synthetic multi-modal instruction databases (MIDBs) for VQA. We then scale up to create OmniVQA-Chat-400K, the largest dataset in the VQA field concurrently. Our focus is on the technical and aesthetic quality dimensions, with abundant in-context instruction data to provide fine-grained VQA knowledge. Additionally, we build the OmniVQA-MOS-20K dataset to enhance the model's quantitative quality rating capabilities. We then introduce a complementary training strategy that effectively leverages the knowledge from datasets for different tasks. Furthermore, we propose the OmniVQA-FG (fine-grain)-Benchmark to evaluate the fine-grained performance of models. Our results demonstrate that our models achieve state-of-the-art performance in both tasks.

Image Quality Assessment (IQA) based on human subjective preferences has undergone extensive research in the past decades. However, with the development of communication protocols, the visual data consumption volume of machines has gradually surpassed that of humans. For machines, the preference depends on downstream tasks such as segmentation and detection, rather than visual appeal. Considering the huge gap between human and machine visual systems, this paper proposes the topic: Image Quality Assessment for Machine Vision for the first time. Specifically, we (1) defined the subjective preferences of machines, including downstream tasks, test models, and evaluation metrics; (2) established the Machine Preference Database (MPD), which contains 2.25M fine-grained annotations and 30k reference/distorted image pair instances; (3) verified the performance of mainstream IQA algorithms on MPD. Experiments show that current IQA metrics are human-centric and cannot accurately characterize machine preferences. We sincerely hope that MPD can promote the evolution of IQA from human to machine preferences. Project page is on: https://github.com/lcysyzxdxc/MPD.

With the rising interest in research on Large Multi-modal Models (LMMs) for video understanding, many studies have emphasized general video comprehension capabilities, neglecting the systematic exploration into video quality understanding. To address this oversight, we introduce Q-Bench-Video in this paper, a new benchmark specifically designed to evaluate LMMs' proficiency in discerning video quality. a) To ensure video source diversity, Q-Bench-Video encompasses videos from natural scenes, AI-generated content (AIGC), and computer graphics (CG). b) Building on the traditional multiple-choice questions format with the Yes-or-No and What-How categories, we include Open-ended questions to better evaluate complex scenarios. Additionally, we incorporate the video pair quality comparison question to enhance comprehensiveness. c) Beyond the traditional Technical, Aesthetic, and Temporal distortions, we have expanded our evaluation aspects to include the dimension of AIGC distortions, which addresses the increasing demand for video generation. Finally, we collect a total of 2,378 question-answer pairs and test them on 12 open-source & 5 proprietary LMMs. Our findings indicate that while LMMs have a foundational understanding of perceptual video quality, their performance remains incomplete and imprecise, with a notable discrepancy compared to the performance of human beings. Through Q-Bench-Video, we seek to catalyze community interest, stimulate further research, and unlock the untapped potential of LMMs to close the gap in video quality understanding.

With the emergence of Multimodal Large Language Models (MLLMs), hundreds of benchmarks have been developed to ensure the reliability of MLLMs in downstream tasks. However, the evaluation mechanism itself may not be reliable. For developers of MLLMs, questions remain about which benchmark to use and whether the test results meet their requirements. Therefore, we propose a critical principle of Information Density, which examines how much insight a benchmark can provide for the development of MLLMs. We characterize it from four key dimensions: (1) Fallacy, (2) Difficulty, (3) Redundancy, (4) Diversity. Through a comprehensive analysis of more than 10,000 samples, we measured the information density of 19 MLLM benchmarks. Experiments show that using the latest benchmarks in testing can provide more insight compared to previous ones, but there is still room for improvement in their information density. We hope this principle can promote the development and application of future MLLM benchmarks. Project page: https://github.com/lcysyzxdxc/bench4bench

The furnishing of multi-modal large language models (MLLMs) has led to the emergence of numerous benchmark studies, particularly those evaluating their perception and understanding capabilities. Among these, understanding image-evoked emotions aims to enhance MLLMs' empathy, with significant applications such as human-machine interaction and advertising recommendations. However, current evaluations of this MLLM capability remain coarse-grained, and a systematic and comprehensive assessment is still lacking. To this end, we introduce EEmo-Bench, a novel benchmark dedicated to the analysis of the evoked emotions in images across diverse content categories. Our core contributions include: 1) Regarding the diversity of the evoked emotions, we adopt an emotion ranking strategy and employ the Valence-Arousal-Dominance (VAD) as emotional attributes for emotional assessment. In line with this methodology, 1,960 images are collected and manually annotated. 2) We design four tasks to evaluate MLLMs' ability to capture the evoked emotions by single images and their associated attributes: Perception, Ranking, Description, and Assessment. Additionally, image-pairwise analysis is introduced to investigate the model's proficiency in performing joint and comparative analysis. In total, we collect 6,773 question-answer pairs and perform a thorough assessment on 19 commonly-used MLLMs. The results indicate that while some proprietary and large-scale open-source MLLMs achieve promising overall performance, the analytical capabilities in certain evaluation dimensions remain suboptimal. Our EEmo-Bench paves the path for further research aimed at enhancing the comprehensive perceiving and understanding capabilities of MLLMs concerning image-evoked emotions, which is crucial for machine-centric emotion perception and understanding. Our code and benchmark datasets are available at https://github.com/workerred/EEmo-Bench.

The advent and proliferation of large multi-modal models (LMMs) have introduced new paradigms to computer vision, transforming various tasks into a unified visual question answering framework. Video Quality Assessment (VQA), a classic field in low-level visual perception, focused initially on quantitative video quality scoring. However, driven by advances in LMMs, it is now progressing toward more holistic visual quality understanding tasks. Recent studies in the image domain have demonstrated that Visual Question Answering (VQA) can markedly enhance low-level visual quality evaluation. Nevertheless, related work has not been explored in the video domain, leaving substantial room for improvement. To address this gap, we introduce the VQA² Instruction Dataset-the first visual question answering instruction dataset that focuses on video quality assessment. This dataset consists of 3 subsets and covers various video types, containing 157,755 instruction question-answer pairs. Then, leveraging this foundation, we present the VQA² series models. The VQA² series models interleave visual and motion tokens to enhance the perception of spatial-temporal quality details in videos. We conduct extensive experiments on video quality scoring and understanding tasks, and results demonstrate that the VQA² series models achieve excellent performance in both tasks. Notably, our final model, the VQA²-Assistant, exceeds the renowned GPT-4o in visual quality understanding tasks while maintaining strong competitiveness in quality scoring tasks. Our work provides a foundation and feasible approach for integrating low-level video quality assessment and understanding with LMMs.

With the rapid advancement of generative models, the realism of AI-generated images has significantly improved, posing critical challenges for verifying digital content authenticity. Current deepfake detection methods often depend on datasets with limited generation models and content diversity that fail to keep pace with the evolving complexity and increasing realism of the AI-generated content. Large multimodal models (LMMs), widely adopted in various vision tasks, have demonstrated strong zero-shot capabilities, yet their potential in deepfake detection remains largely unexplored. To bridge this gap, we present DFBench, a large-scale DeepFake Benchmark featuring (i) broad diversity, including 540,000 images across real, AI-edited, and AI-generated content, (ii) latest model, the fake images are generated by 12 state-of-the-art generation models, and (iii) bidirectional benchmarking and evaluating for both the detection accuracy of deepfake detectors and the evasion capability of generative models. Based on DFBench, we propose MoA-DF, Mixture of Agents for DeepFake detection, leveraging a combined probability strategy from multiple LMMs. MoA-DF achieves state-of-the-art performance, further proving the effectiveness of leveraging LMMs for deepfake detection. Database and codes are publicly available at https://github.com/IntMeGroup/DFBench.

The rapid evolution of Multi-modality Large Language Models (MLLMs) is driving significant advancements in visual understanding and generation. Nevertheless, a comprehensive assessment of their capabilities, concerning the fine-grained physical principles especially in geometric optics, remains underexplored. To address this gap, we introduce GOBench, the first benchmark to systematically evaluate MLLMs' ability across two tasks: 1) Generating Optically Authentic Imagery and 2) Understanding Underlying Optical Phenomena. We curate high-quality prompts of geometric optical scenarios and use MLLMs to construct the GOBench-Gen-1k dataset. We then organize subjective experiments to assess the generated imagery based on Optical Authenticity, Aesthetic Quality, and Instruction Fidelity, revealing MLLMs' generation flaws that violate optical principles. For the understanding task, we apply crafted evaluation instructions to test the optical understanding ability of eleven prominent MLLMs. The experimental results demonstrate that current models face significant challenges in both optical generation and understanding. The top-performing generative model, GPT-4o-Image, cannot perfectly complete all generation tasks, and the best-performing MLLM model, Gemini-2.5Pro, attains a mere 37.35% accuracy in optical understanding. Database and codes are publicly available at: https://github.com/aiben-ch/GOBench.

With the rapid development of streaming media technology, the real-time streaming video Quality of Experience (QoE) assessment has become an important objective for creating new Adaptive Bitrate (ABR) algorithms. The QoE prediction on the client side is challenging considering the sophisticated perception mechanisms of humans, especially the human attention behaviors over time. To address this issue, we propose a learnable model based on the dual-stage attention mechanism to precisely predict continuous QoE which is not covered by most of the current related works. Given the close relationship between the continuous and overall QoE, we use a unified framework to predict these two indices. We have conducted comparison experiments on 6 open datasets, and our model shows superior performance.

The imaging quality of automotive cameras is crucial in complex driving environments. Therefore, it is essential to conduct subjective experiments that can realistically reflect drivers’ evaluation of the imaging quality of automotive cameras in real traffic scenarios. To accurately assess the imaging quality of automotive cameras, this paper proposes a no-reference quality assessment method with quality scores that are highly consistent with human subjective perception. Initially, this study constructs a new image quality assessment dataset and then obtains the subjective scores of image quality through subjective experiments. The dataset is constructed by using a variety of realistic props to simulate scene elements that might be captured by an automotive camera and are captured using a wide range of cameras with different sensor types, lens focus, and viewing angles, resulting in a dataset of diverse images. The objective quality assessment method proposed in this paper consists of an object detection network and a multi-branch quality evaluation network. The object detection network is responsible for identifying and classifying scene elements, while the multi-branch quality evaluation network performs feature extraction and score regression on various types of elements to effectively evaluate the imaging quality of the automotive cameras. In the experiments, this no-reference quality assessment method is tested on our built dataset, and the results show that the proposed method exhibits the best performance compared with the state-of-the-art image quality assessment methods.

With the rapid development of video-on-demand (VOD) and real-time streaming video technologies, the accurate objective assessment of streaming video Quality of Experience (QoE) has become a focal point for optimizing streaming-related technologies. However, due to the inherent transmission distortions caused by poor Quality of Service (QoS) conditions in streaming videos, such as intermittent stalling, rebuffering, and drastic changes in video sharpness due to bitrate fluctuations, evaluating streaming video QoE presents numerous challenges. This paper introduces a large and diverse in-the-wild streaming video QoE evaluation dataset - the SJLIVE-1k dataset. This work addresses the limitations of corresponding datasets, which lack in-the-wild video sequences under real network conditions and whose amount of video content is insufficient. Furthermore, we propose an end-to-end objective QoE evaluation strategy that extracts video content and QoS features from the video itself without using any extra information. By implementing self-supervised contrastive learning as the "reminder" to bridge the gap between the different types of features, our approach achieves state-of-the-art results across three datasets. Our proposed dataset will be released to facilitate further research.

With the rapid development of streaming media technology, the Quality of Experience (QoE) of streaming videos becomes crucial to optimize the video compression and transmission algorithms, such as adaptive bitrate (ABR). However, the complexity of human perceptual mechanisms, particularly in relation to temporal distortions, poses substantial challenges to effective QoE monitoring. In recent years, many efforts in video quality assessment (VQA) and video QoE evaluation have highlighted the influence of a broad spectrum of features—from Quality of Service (QoS) metrics to video content understanding—on viewer experience. On this basis, we believe that there is also a dynamic relationship among these features varying with the broadcasting content. Furthermore, research indicates a significant correlation between real-time and retrospective assessments of QoE for individual videos. In response to these insights, we introduce a novel approach leveraging a unified learnable network that incorporates dual-stage attention, the temporal and cross-feature attention, to accurately predict both continuous and overall QoE for streaming videos. The results of experiments conducted on several publicly available databases demonstrate the superiority of our proposed method over the state-of-the-art metrics.