Abstract: By analyzing the composition and functionality of security inspection intelligent centralized image judgment system, setting schemes are proposed. Independent centers for centralized image judgment are established in units of urban rail transit FAO (fully automatic operation) line, station and centralized station. Meanwhile, intelligent algorithm based on deep learning assists in the judgment of security inspection images. By combining intelligent and centralized image judgment methods, the real intelligentized inspection is achieved. For each line, an appropriate security inspection intelligent centralized image judgment scheme can be selected according to the specific passenger flow volume, mileage and number of stations. Taking Nanning Rail Transit Line 5 phase I project as an example, the economic benefit analysis shows that the security inspection intelligent centralized image judgment system can effectively reduce labor cost and realize both personnel reduction and efficiency improvement in the image judgment process.