Pear Blossom Counting System with Drones Using YOLO and Deep SORT

In this paper, we propose a pear blossom counting system with drones using YOLO (You Only Look Once) and Deep SORT (Simple Online and Realtime Tracking) for pollination system. To establish the pollination system, it is necessary to understand the blossoming status of pear blossoms. In our proposed system, a drone is flown to capture movies among pear trees, and the number of pear blossoms and buds are counted from the images using a object detection method, i.e, YOLO. If the number of pear blossoms and buds can be counted, it is possible to estimate the best time for pollination. Moreover, to estimate the number of blossoms, we use a tracking algorithm, i.e., Deep SORT. Through experiments using movies taken during the blossoming period, we show that the proposed method can count the number of blossoms and buds.