Long-Range Drone Detection Dataset

Abstract

For the safe and efficient deployment of unmanned aerial vehicles (UAVs) in complex urban landscapes, robust collision avoidance mechanisms are imperative. Although several methodologies exist for drone detection, current solutions are suboptimal for long-range detection, primarily due to the scarcity of comprehensive training datasets. In this paper, we present a novel long-range drone detection dataset, encompassing a set of different UAV types, flight patterns, and environmental conditions. Utilizing this dataset, we trained a state-of-the-art YOLO object detection algorithm, demonstrating the ability to identify drones at distances up to 60 meters with a high mean average precision (mAP). Extensive real-world tests affirm the efficacy of our approach, achieving a detection accuracy exceeding 75%. This dataset and the accompanying machine learning model contribute a significant advancement in the realm of long-range drone detection, particularly well-suited for urban deployments. For access to the complete Long-Range Drone Detection Dataset (LRDD), please visit https://research.coe.drexel.edu/ece/imaple/long-range-drone-detection-dataset/.