Vehicle Recognition Based on the Carrier-Free UWBR Using ViT Model and 2D-Echo Feature Images

In this letter, we propose a novel approach for vehicle object recognition based on a carrier-free wideband radar (UWBR) using 2D echo images and the Vision Transformer (ViT) model. The input image is generated by assembling 1D echo signals under multiple azimuth angles into 2D echo feature images. The performance of the proposed method is evaluated and compared with advanced Convolutional Neural Networks (CNNs) models such as LeNet, VGG16, and ResNet50. Experimental results demonstrate that the Vision Transformer model can achieve better performance in vehicle object recognition tasks, surpassing some advanced CNNs.