Development of a Technique for Predicting the Harvest Maturity of Sweet Corn Using Object Detection AI

Abstract

　A technique was developed for predicting the harvest maturity of sweet corn using object detection AI. The estimated harvest period is output by inputting aerial images, location of the sweet corn field, variety and the date of aerial image taken. The object detection model “Tassel detection AI” and the “Agro-Meteorological Grid Square Data, NARO” were used for the analysis of aerial images and the estimation of harvesting maturity. The prediction accuracy of the tool, that overlapped between the harvest period and the predicted harvest period, was approximately 70 %. This tool can predict the best time to harvest approximately one month before harvest. This capacity contributes to the efficient planning of harvest operations by this tool users.