Remote Sensing-based Automated Civil Aircraft Detection using YOLOv8 Architecture

Detection of aircraft from satellite images is a critical activity in air traffic observation, airport operations, and air safety. Conventionally, the detection of aircraft is based on manual examination or traditional machine learning methods, which are challenged with coping with variation in aircraft shape, orientation, and weather conditions. In this paper, we introduce a deep learning approach based on the YOLOv8 object detection model to identify civil aircraft from high-resolution satellite images. Although several existing models like Faster R-CNN, SSD, and previous YOLO versions have been used for object detection tasks, YOLOv8 offers practical improvements. It simplifies the detection pipeline by removing anchor boxes, uses a cleaner architecture, and provides better training stability. The current work targets testing the model using accuracy metrics like mean Average Precision (mAP) and Intersection over Union (IoU) and comparisons with other deep learning models. The model has achieved an accuracy of 91% and false positive rate of 0.31%. The objective of this research is to make a contribution to automated satellite image-based aircraft detection, with possible implications in real-time aviation surveillance and operational planning.