Despite several efforts to spot unmanned aerial vehicles (UAVs) using computer vision, no benchmark dataset is available for the research community to use and compare results. This study presents three single-shot UAV spotter frameworks based on YOLOv4, YOLOv5, and DETR architectures, along with an open-source dataset and performance metrics to address this problem.
(DOI: 10.1109/ICECCE52056.2021.9514124)