Classification of drones and birds using convolutional neural networks applied to radar micro-Doppler spectrogram images

Abstract

This study presents a convolutional neural network-based drone classification method. The primary criterion for ahigh-fidelity neural network-based classification is a real dataset of large size and diversity for training. The first goal of the studywas to create a large database of micro-Doppler spectrogram images of in-flight drones and birds. Two separate datasets withthe same images have been created, one with RGB images and others with greyscale images. The RGB dataset was used forGoogLeNet architecture-based training. The greyscale dataset was used for training with a series of architecture developedduring this study. Each dataset was further divided into two categories, one with four classes (drone, bird, clutter and noise) andthe other with two classes (drone and non-drone). During training, 20% of the dataset has been used as a validation set. Afterthe completion of training, the models were tested with previously unseen and unlabelled sets of data. The validation and testingaccuracy for the developed series network have been found to be 99.6 and 94.4%, respectively, for four classes and 99.3 and98.3%, respectively, for two classes. The GoogLenet based model showed both validation and testing accuracies to be around99% for all the cases.