Multi-task Feature Enhancement Network for No-Reference Image Quality Assessment (NR-IQA)
This research paper introduces a novel multi-task deep learning framework for no-reference image quality assessment (NR-IQA) that outperforms existing methods by leveraging high-frequency image information and a distortion-aware network.
ResNet-L2: A Novel and Efficient Metric for Evaluating Generative Models and Image Quality in Histopathology Images
This paper introduces ResNet-L2 (RL2), a novel metric for evaluating the quality of generative models and images in histopathology, addressing the limitations of traditional metrics like FID, especially in data-scarce scenarios.
인공위성 영상의 선명도 평가를 위한 새로운 무참조 영상 화질 지표
이 연구 논문에서는 인공위성 영상에서 선명도를 객관적으로 평가하기 위해 새롭게 개발된 무참조 영상 화질 지표를 소개합니다.
고품질 압축 이미지에 대한 세분화된 주관적 화질 평가 방법론 및 데이터셋
고품질 압축 이미지의 미묘한 화질 차이를 평가하기 위해 부스팅 기법을 사용한 삼중 자극 비교 방법론과 JND 단위의 세분화된 품질 척도를 제시하고, 대규모 크라우드소싱을 통해 구축된 데이터셋과 분석 결과를 소개한다.
A New Methodology for Evaluating the Quality of High-Fidelity Compressed Images
This research paper introduces a novel subjective quality assessment methodology for high-fidelity compressed images, employing boosted and plain triplet comparisons to achieve a fine-grained quality scale in Just Noticeable Difference (JND) units, offering more informative results for practical applications than traditional methods.
基於屬性感知預訓練的通用圖像質量特徵提取器：ATTIQA
針對無參考圖像質量評估 (NR-IQA) 中數據集大小有限的問題，本文提出了一種名為 ATTIQA 的新型預訓練框架，該框架利用屬性感知預訓練從大型數據集中提取與質量相關的知識，從而構建用於 IQA 的通用表示，並在多個 IQA 數據集上實現了最先進的性能。
ATTIQA: A Novel Pretraining Framework for Generalizable No-Reference Image Quality Assessment Leveraging Attribute-Aware Pretraining with Vision Language Models
This paper introduces ATTIQA, a novel pretraining framework for No-Reference Image Quality Assessment (NR-IQA) that leverages attribute-aware pretraining with Vision Language Models (VLMs) to achieve state-of-the-art performance and superior generalization capabilities.
Dog-IQA: A Training-Free, Standard-Guided, and Mix-Grained Image Quality Assessment Method Using Multimodal Large Language Models
This paper introduces Dog-IQA, a novel training-free method for image quality assessment (IQA) that leverages the capabilities of pre-trained multimodal large language models (MLLMs) and segmentation models to achieve state-of-the-art performance in zero-shot settings.
© 2024 by Linnk AI