Improving the YOLO v4 algorithm applied to the powdery mildew and downy mildew monitor system on cucumber plants

Abstract

The technology of artificial intelligence, robotics, and IoT is developing very strongly. The application of these technologies to serve in agricultural production is very necessary to reduce labor and increase crop productivity. Monitoring and warning crop diseases promptly will help to treat diseases faster, save financial resources and increase crop yields. By combining artificial intelligence algorithm with automatic data acquisition and information transmission system, we have designed a robot to monitor and warn against powdery mildew and downy mildew on the cucumber plants. Our improved YOLO v4-tiny algorithm gave an accuracy of 83.26% which is 0.22% higher than the original YOLO v4-tiny version; the system initially tested with Raspberry PI and Camera PI 5MB hardware gave good results. The system will continue to study to upgrade the hardware, as well as add the dataset so that it can be put into cucumber production in the net house.