Detection and Recognition of Drones using Deep Convolution Neural Networks

Drones are gaining popularity in various fields, including delivering products, disaster management, entertainment, airport security, and more. As a result, the odds of a drone being utilized for malevolent purposes are increasing, prompting an unsettling concern about physical infrastructure security, safety, and monitoring at airports. There have been numerous allegations in recent years of unauthorized usage of various types of drones at airports, causing airline operations to be disrupted. To solve this issue, this study proposed two deep learning-based approaches named YOLOv4 and Faster RCNN. The offered methodologies were compared to find the optimal model. Drones are also frequently mistaken for birds due to their morphological and behavioral similarities. Not only can the proposed algorithms identify the presence of drones in a provided image, but they can also detect the presence of birds and point to their positions. The proposed YOLOv4 model obtained better results compared to Faster RCNN, with mAP values of 75% and 72%, respectively.