Title: Improving the Performance of YOLO-based Detection Algorithms for Small Object Detection in UAV-taken Images Defense
Öykü Şahin, M.Sc. Student in Computer Engineering
Advisor: Prof. Dr. İbrahim Körpeoğlu
Co-Advisor: Prof. Dr. Sedat Özer
Member: Prof. Dr. Özgür Ulusoy
Asst. Prof. Dr. Mustafa Furkan Kıraç (Özyeğin University)
The Seminar will be on Tuesday, January 3, 2023, 16:00
This is an online seminar. To request the event link, please send a message to department.
Recent advances in computer vision yield emerging novel applications for camera-equipped unmanned aerial vehicles such as object detection. The accuracy of object detection solutions running on images taken from Unmanned Aerial Vehicles (UAVs) is limited when compared with the performance of the object detection solutions for ground-taken images. Existing object detection solutions demonstrate lower performance on aerial datasets because of the reasons coming from the nature of the UAVs. These reasons are: (i) the lack of large drone datasets with different types of objects, (ii) the larger variance in both scale and orientation in drone images, and (iii) the difference in shape and texture of the features between the ground and the aerial images. Due to these reasons, YOLO-based models, a popular family of one-stage object detectors, perform lower in UAV-based applications. In this thesis, improved YOLO models: YOLODrone and YOLODrone+ are introduced for detecting objects in drone images. The performance of the models were tested on VisDrone2019 and SkyDataV1 datasets and improved results were reported when compared to the original YOLOv3 and YOLOv5 models.