YOLO-Based Approach for Intelligent Apple Crop Health Assessment

The escalating demands of human reproduction and consumption have placed immense pressure on agricultural companies to enhance production speed and size. However, rapid growth in agricultural operations has led to an increase in human errors, affecting the quality and safety of agricultural products. To mitigate these repercussions, agro-businesses are exploring the potential of artificial intelligence (AI) technologies, specifically computer vision, to detect anomalies and assess the health of agricultural produce. This research paper focuses on utilizing YOLO-based object detection to evaluate the health of apples by detecting common diseases and disorders. The proposed methodology involves training three versions of YOLO models (YOLOv5s, YOLOv5m, and YOLOv7) using different optimization algorithms. The performance results show that YOLOv5m achieved the highest mean average precision (mAP) score of 89% using SGD with Nesterov momentum.