Foreign Object Detection and Classification using AI and ML for Radio Images

Ultra-wideband (UWB) radar has many applications in surveillance and security as it can perform contactless Electromagnetics (EM) detection of concealed objects and Improvised explosive devices (IEDs). However, automatic object detection and classification of radio images is a slightly difficult task due to the high level of noise present in sub-surface media with large clutter diversity. Hence, in recent days, artificial intelligence (AI) and machine learning (ML) based radar imaging technologies are emerging prominently. In this paper, ML algorithms such as Support vector machine (SVM), Random forest, Convolutional neural network (CNN), and VGG16 were used for the classification of various hidden targets and IEDs. It's found that the SVM classifier has achieved a reasonable accuracy of 84.07% compared to other classifiers. To further improve the accuracy, You only look once (YOLOv3) network was implemented for object detection and training purposes. The proposed YOLOv3 model has achieved mean Average Precision (mAP) accuracy of 99.6%. Therefore, the proposed research work has a unique capability for the detection and classification of various foreign objects for both civil and military applications.