AERIAL OBJECT DETECTION SYSTEM WITH DEEP LEARNING

Abstract

Nowadays micro/mini drones become highly accessible to the person from all walks of life. The statement mentioned above poses enormous safety hazards and regulatory challenges. Due to the smaller radar reflection cross-sectional area of unauthorized drones, difficult to detect by radio detection system, which may interfere the normal takeoff and landing progress of aircraft or leak the location information of facilities. In recent years, deep learning methods have made good progress in the field of small object detection. Therefore, we suggest, in this paper, a drone detection method that integrates deep learning-based classification and localization tasks. Using YOLO v8(You Only Look Once Version 8), deep learning neural network, and adjusting its architecture and parameters to better adapt to small object detection such as micro/mini drones. In addition, to train the neural network model in this article to classify detected aerial aims, we selected a multi class flying object dataset that includes birds, drones, helicopters, and fixed wing aircraft, among which some may be potential threats.