Advances in Sensor-Based and Machine Learning Techniques for Automated Tomato Sorting and Grading: A Review

Abstract

The increasing demand for high-quality tomatoes necessitates the implementation of efficient sorting and grading systems to meet market standards. Traditional manual sorting methods are labour-intensive and susceptible to inaccuracies, thereby highlighting the need for automated solutions. This review examines advancements in sensor-based technologies and machine learning algorithms for automated tomato sorting and grading. A variety of methodologies, including the utilization of RGB colour sensors, convolutional neural networks (CNNs), and hybrid models, were evaluated for their efficacy in classifying tomatoes based on size, colour, ripeness, and defects. The review examines systems that achieve classification accuracies of up to 97.5%, with sorting capacities of up to 2807 tomatoes per hour. Challenges such as variability in performance across different tomato varieties and high implementation costs were identified as significant barriers to widespread adoption. The findings suggest that further research should concentrate on enhancing the scalability, adaptability, and affordability of these systems, with potential improvements through the integration of emerging technologies such as artificial intelligence (AI) and the Internet of Things (IoT). Overall, automated sorting systems hold considerable promise for improving quality control in tomato production and ensuring a consistent supply for global markets.