Drone Model Classification Using Convolutional Neural Network Trained on Synthetic Data

Mariusz Wisniewski; Zeeshan A Rana; Ivan Petrunin

doi:10.3390/jimaging8080218

Drone Model Classification Using Convolutional Neural Network Trained on Synthetic Data

J Imaging. 2022 Aug 12;8(8):218. doi: 10.3390/jimaging8080218.

Authors

Mariusz Wisniewski¹, Zeeshan A Rana¹, Ivan Petrunin¹

Affiliation

¹ Digital Aviation Research and Technology Centre (DARTeC), Cranfield University, Cranfield MK43 0FQ, UK.

Abstract

We present a convolutional neural network (CNN) that identifies drone models in real-life videos. The neural network is trained on synthetic images and tested on a real-life dataset of drone videos. To create the training and validation datasets, we show a method of generating synthetic drone images. Domain randomization is used to vary the simulation parameters such as model textures, background images, and orientation. Three common drone models are classified: DJI Phantom, DJI Mavic, and DJI Inspire. To test the performance of the neural network model, Anti-UAV, a real-life dataset of flying drones is used. The proposed method reduces the time-cost associated with manually labelling drones, and we prove that it is transferable to real-life videos. The CNN achieves an overall accuracy of 92.4%, a precision of 88.8%, a recall of 88.6%, and an f1 score of 88.7% when tested on the real-life dataset.

Keywords: airport security; artificial intelligence; convolutional neural network; domain randomization; drone classification; drone detection; drone identification; drones; synthetic data; synthetic images; unmanned aerial vehicles.

Grants and funding

This project is funded by the Department for Transport courtesy of the Future Aviation Security Solutions (FASS) programme, in collaboration with Aveillant Ltd. and Autonomous Devices Ltd.