Approach and Analysis of Yolov4 Algorithm for Rice Diseases Detection at Different Drone Image Acquisition Distances

Abstract

Rice production plays an important role in people's lives globally because rice is the most widely consumed staple food for over half of the world's human population. Unfortunately, the rice plants are prone to pests and diseases which may result in a decrease in the rice production. Thus, early and accurate detection of rice diseases is needed. This paper discusses a method for detecting rice diseases called You Only Look Once (YOLO) object detection algorithm version 4. A drone camera was used to acquire the images from four different distances, namely 2 meters, 5 meters, 10 meters and 20 meters. This approach aims to detect the presence of pests to be faster, more accurate and more precise. The precision results at each image capture from the different distances were 46.8%, 48%, 65% and 77.3% with the average loss value of 6.52, 0.54, 1.16 and 2.73