Computer vision in aquaculture: transforming fish freshness monitoring

Fish assessment plays a crucial role in ensuring the freshness and safety of fish products, thereby enhancing consumer satisfaction and improving supply chain management. The rapid advancement of computer vision (CV) and image processing technologies offers a promising noninvasive solution for evaluating fish freshness, enabling the detection and classification of freshness indicators efficiently. This review presents an overview of the application of CV techniques in fish freshness assessment, focusing on advancements over the past decade. The fish freshness evaluation task is categorized into three key aspects such as freshness indicators, texture features, and biochemical changes. Traditional methods as well as deep learning approaches in CV are summarized, highlighting their respective applications for assessing these freshness aspects. Furthermore, commonly used algorithmic models in fish freshness detection are comprehensively introduced. Additionally, the review discusses the integration of advanced imaging technologies, such as hyperspectral and thermal imaging, which provide detailed insights into biochemical and microbial alterations that traditional imaging techniques fail to detect. Finally, the review identifies several challenges in fish freshness assessment, proposing feasible strategies to address these issues and underlining the importance of improving models to integrate data from various sources.