Vehicle automatic driving technology can effectively improve the safety performance of vehicle driving. This research is aimed at the needs of vehicle automatic driving. Combined with vehicle perception technology, a better target recognition algorithm is proposed. By comparing the recognition and recall rate of ION algorithm, HYPERNET algorithm, R-CNN (Regions with CNN features) algorithm, and multi-strategy region proposal network (MSRPN) algorithm, it can be seen that MSPRN algorithm has better algorithm performance and is suitable for target detection and recognition in vehicle automatic driving.