Determination of Fruit Health Status and Yield with Unmanned Aerial Vehicle

In this study, it is aimed to determine the number of reference fruits and health status (sturdy, rotten, mottled, non-spotted) by using real-time image or recorded video taken from the autonomous Unmanned Aerial Vehicle (UAV) camera in orchards. In the determinations made by using image processing techniques, sturdy-rotten and mottled-speckless distinction are made for oranges and apricots, respectively. These distinction and determination processes are carried out using highly trained classifiers. Three types of multi-trained classifiers performance have been compared and a highly trained classifier which has high performance has been preferred for object detection. The accuracy of the Haar, local binary pattern (LBP), and histogram of oriented gradients (HOG) classifiers are compared in Python using the open source computer vision library. It has been shown experimentally that Haar classifier achieves high performance in determining realtime reference fruit health status and yield.