Department of Computer Engineering
One-Stage Oriented Object Detection in Remote Sensing Images
(Supervisor: Prof. Dr. Selim Aksoy )
Computer Engineering Department
Abstract: Advances in technology resulted in enormous amount of information collected from high resolution satellites and aircraft sensors. These high resolution images obtained by the said platforms enabled humans to understand the Earth better. Earlier work in the field of remote sensing image analysis involved image classification that studies the understanding of scenes, terrain, and vast fields. Recently object detection and image segmentation on very high resolution electro-optical images have become popular. We focus on object detection in a large-scale object detection dataset that contains 15 different object classes with varying sizes. First, we use a two-stage object detector method to obtain results and then study one-stage object detection. Our one-stage object detectors are based on YOLO architecture. Horizontal and oriented bounding box labels are used to obtain results to show that oriented object detection setting has a surpassing performance over axis-aligned object detection setting in general and also to show that one-stage object detection can have competitive results compared to two-stage object detection methods. Our oriented object detection approach converts angle regression into a form of classification and is easily applicable into the one-stage detector.
DATE: 31 December 2021, Friday @ 13:30 Zoom