Abstract: In order to solve the problem that the obstacle detection of transport vehicles in roadway is easily affected by lighting conditions, an obstacle detection method based on image enhancement is proposed. Firstly, the VOC data set format was used to produce the obstacle data set in the process of underground vehicle driving, and then MSR (Multi Scale Retinex) algorithm was used to enhance the low-illumination image collected in the underground. By improving the CenterNet network, the ResNetBN-18 lightweight network was designed, and the data set was trained based on the improved CenterNet target detection algorithm. Finally, the accurate detection of vehicles road obstacle in underground roadway was realized. The experimental results show that the improved detection model maintains the high real-time performance of the original network, and the detection accuracy is improved by 10.1%, the frame rate increased by 6.7%.