EZM-AI: A Yolov5 Machine Vision Inference Approach of the Philippine Corn Leaf Diseases Detection System
Yolanda D. Austria, Maria Concepcion A. Mirabueno, Dylan Josh Domingo Lopez, Dexter James L. Cuaresma, Jonel R. Macalisang, Cherry D. Casuat
Abstract
The Philippines is an agricultural country, and one of the issues in today's farming environment is the prevalence and exacerbation of diseases caused by fungus, which impact the overall quality of the produced or harvested crop. This study focuses on a corn field, especially the top three corn crop diseases in the Philippines, which are corn rust, leaf blight, and grey leaf spot. The YOLO V5 architecture was used to identify corn crop diseases. After training, the result had an mAP score of 0.97. The model also achieved 100 percent testing accuracy and detection accuracy ranging from 98.90 percent to 99.43 percent. The accuracy of training, testing, and validation were promising, and it could be implemented into the device to solve the issue of detecting corn leaf diseases.