Artificial Intelligence and Deep Learning-Based System for Agri-Food Quality and Safety Detection

Abstract

Deep Learning (DL) has emerged as a highly effective technique for analyzing large volumes of data across various domains, including image processing, speech recognition, and pattern recognition. Recently, DL has also found applications in the field of food science and engineering, a relatively novel area of research. This paper provides a concise introduction to DL and delves into the architecture of a typical Convolution Neural Network (CNN) structure, as well as AI and IoT (Internet of Things) data training methodologies. Our research involved an extensive review of studies that utilized DL as a computational approach to address food-related challenges, such as food recognition, calorie computation, and safety detection of various food types like fruits, potatoes, meats, and aquatic products, as well as food supply chain management and food borne illness detection. Each study examined different problems, datasets, preprocessing techniques, network architectures, and evaluation metrics, comparing their results with alternative solutions. Furthermore, we explored the role of big data in the field of food quality assurance, uncovering compelling trends. Based on our analysis, DL consistently outperforms other approaches, including manual feature extractors and traditional machine learning algorithms. The findings highlight the tremendous potential of DL as a promising technology for food safety inspections and related applications in the food industry