Deep Learning Based Object Detection and Recognition of Unmanned Aerial Vehicles

In this study, the methods of deep learning-based detection and recognition of the threats, evaluated in terms of military and defense industry, by unmanned aerial vehicles (UAV) are presented. In the proposed approach, firstly, the training for machine learning on the objects is carried out using convolutional neural networks, which is one of the deep learning algorithms. By choosing the Faster-RCNN and YoloV4 architectures of the deep learning method, it is aimed to compare the achievements of the accuracy in the training process. In order to be used in the training and testing stages of the recommended methods, data sets containing images selected from different weather, land conditions and different time periods of the day are determined. The model for the detection and recognition of the threatening elements is trained, using 2595 images. The method of detecting and recognizing the objects is tested with military operation images and records taken by the UAVs. While an accuracy rate of 93% has been achieved in the Faster-RCNN architecture in object detection and recognition, this rate has been observed as 88% in the YoloV4 architecture.