retIQA

Traditional IQA methods, often rooted in handcrafted features and statistical metrics, have demonstrated limitations in capturing the complexities of human perception and preferences. To address these challenges, the retrieval approach has emerged as a promising avenue, synergizing the power of deep learning with information retrieval techniques. This work presents a comprehensive exploration of the retrieval-augmented deep learning paradigm for IQA, proposing a novel framework that leverages both image content and contextual information to achieve improved and perceptually aligned quality assessment. Through extensive experimentation and comparative analysis, we demonstrate the efficacy of this approach in enhancing the accuracy and reliability of IQA models, thus contributing to the evolution of image quality assessment in the era of data-driven paradigms.