Drone Dataset: Amateur Unmanned Air Vehicle Detection

This dataset is prepared for our 2019 year "Amateur Drone Detection and Tracking" project. There are more than 4000 amateur drone pictures in the dataset, which is usually trained with amateur (like dji phantom) drones. In addition, the dataset contains non-drone, drone-like "negative" objects. This dataset was used with Yolov2-tiny, Yolov3-voc versions. Generally suitable for working with Yolo architecture and darknet framework.