Traffic Target Recognition for 5G Millimeter-Wave Base Stations Based on Deep CNNs
Qinglong He, Jian Wang, Bodong Wen, Ance Xu, Muguang Wang · 2025
In this paper, a 5G millimeter-wave base station traffic object recognition method based on deep convolutional neural network (CNN) and multimodal data fusion is proposed. A target recognition model based on Range-Doppler map data provided by the base station is designed, and the sensitivity of the network to target radar features is improved by introducing deformable convolution. Experimental results show that a classification accuracy of 92.38% is achieved on a three-classification dataset, which validates its effectiveness in the traffic target recognition task. This method not only improves the recognition accuracy, but also excels in computational efficiency and can provide a robust and efficient solution for object detection in intelligent transportation.