UAV Data Processing using the YOLOv5 Neural Network

This study uses the YOLOv5 method to detect and classify small objects for unmanned aerial vehicle data. The work uses datasets with different scales to train the detector for pattern recognition from different heights. The YOLOv5 detector trained on the formalized structure data is further capable of recognizing objects in the images at high speed. The recommendation part of the paper describes options for improving the existing algorithm to solve the problem of recognizing small objects from unmanned aerial vehicles more accurately.