DETECTING AIRPLANES IN UAV IMAGES USING DEEP LEARNING MODEL
Computer and artificial intelligence technology has been researched and developed since the 1950s. In the 1980s, research efforts in the field of deep learning, a subfield of machine learning, intensified. Currently, deep learning has provided solutions to many problems. Object detection technology has made significant progress with the use of algorithms such as You Only Look Once (YOLO), Open CV, etc., which enable instant detection of multiple objects in an image. Recently, UAVs have become critically important for countries. The use of UAVs in important studies is increasing day by day due to their areas of use and tasks. Real-time operation and the ability to take images of the desired quality according to camera capabilities are a data source for deep learning studies. This study aims to detect civil and military aircraft from UAV images using different versions of YOLO deep learning algorithms. For this purpose, a dataset was first created to train the YOLO deep learning model. Python-based programming language was used as software for training the model, and different libraries were used. Images containing different types and categories of aircraft were first taught to the created model. Then, for verification purposes, images were given to the model with the relevant codes and the ability to detect the aircraft object was measured. During the tests, the parameters required for the selection of the appropriate model were changed and their effects on the system were examined. Considering all these tests, it is estimated that the use of UAVs will increase and, in parallel, object detection with images obtained from UAVs will have critical importance in the fields of military, agriculture, health and autonomous technology.
Index Terms– AI, UAV, artificial intelligence, deep learning, object detection, YOLO
