A Systematic Literature Review on Explainable AI-Based Fish Quality Assessment Using Deep Learning

Fish quality assessment is essential for ensuring food safety, maintaining consumer confidence, and supporting the global seafood industry. Recent advances in Artificial Intelligence (AI), particularly deep learning and Explainable Artificial Intelligence (XAI), have enabled accurate, rapid, and non-destructive evaluation of fish freshness and quality. This systematic literature review aims to provide a comprehensive analysis of AI-based fish quality assessment techniques, with a particular focus on explainable deep learning models and their applications in seafood inspection. The review was conducted following the PRISMA 2020 guidelines and included publications from 2019 to 2026 retrieved from Scopus, Web of Science, IEEE Xplore, ScienceDirect, SpringerLink, and the ACM Digital Library. Following the screening and eligibility process, 220 representative studies were included for qualitative synthesis. The reviewed literature covers Convolutional Neural Networks (CNNs), transfer learning, Vision Transformers (ViTs), hyperspectral imaging, thermal imaging, multimodal sensing, and Explainable AI techniques, including Grad-CAM, LIME, SHAP, saliency maps, and attention visualization. Comparative analysis indicates that deep learning models consistently achieve classification accuracies exceeding 90–95%, while XAI methods significantly improve model transparency, interpretability, and user trust in industrial decision-making. However, challenges such as limited benchmark datasets, poor cross-species generalization, computational complexity, and the absence of standardized explainability evaluation frameworks continue to hinder widespread industrial adoption. This review identifies current research trends, summarizes existing datasets and evaluation metrics, highlights critical research gaps, and proposes future directions, including Edge AI, federated learning, digital twins, multisensor fusion, and trustworthy XAI, to support the development of intelligent, transparent, and reliable fish quality assessment systems.