Computer Vision for Food Quality Assessment: Advances and Challenges

Abstract

One of the transformative technologies in food quality assessment provided by computer vision is an automated, precise and efficient food product evaluation. In this review, the progress, the difficult issues, and the future way of computer vision applied in food quality control are discussed. Advances in imaging technology, artificial intelligence, and deep learning have improved food inspection accuracy to real time defect detection, ripeness estimation and contamination detection. Feature extraction and classification using hyper spectral imaging and neural networks, Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs) have been improved to design robust and efficient schemes for food quality assessment. Although these breakthroughs have been made, problems like food variability, dependence on large annotated databases, high implementation costs, and real time processing limitations hold back the common use. The complexity of the vision system integration in the industrial food production still remains a concern, especially for small or medium size enterprises. Future research aimed at integrating IoT and edge computing for real time monitoring, explainable AI for transparent decision making, and multimodal data fusion for accurate fusion would address the above mentioned challenges. Moreover, the creation of sustainable and low cost computer vision solutions will be crucial in ensuring availability of these solutions in different food industry sectors.