Multi-target Detection and Classification for Intelligent Vehicle Based on Deep Learning

A multi-target detection and classification method was presented for intelligent vehicle. This method is based on deep learning. Multi-target detection and classification under the traffic road scene is hard to realize since the complexity of road traffic conditions and the diversity of demand. By finding a suitable deep learning algorithm, each detection and recognition task is completed. Based on You Only Look Once (YOLO) v5 algorithm, the task of road multi-target detection is completed. Based on Deep Simple Online and Real Time Tracking (Deep SORT) algorithm, the target tracking of YOLOv5 detection results is realized. Based on Multi-Task Convolutional neural network (MTCNN), the license plate recognition is realized. The built system is adopted to guarantee target detection and tracking of input video. The experiment results are presented and the results show the effectiveness and high accuracy of the target detection and tracking.