Using X-ray or γ-ray to inspect vehicles is an effective way to find threats such as stowaways and contrabands hidden inside. Popularly used vehicle inspection devices at present require passengers to get off the vehicle, and then the vehicle is dragged by a conveyor to move through the inspection device. One disadvantage of such devices is that inspecting efficiency is very low. In this paper, a drive-through mode passenger vehicle inspecting system is introduced, with which passengers in a vehicle need not get off and efficiency is greatly improved. Patented high-pressure gas ionization chamber is used as radiation detector to collect the low radiation signal. A deep-learning study algorithm is adopted to realize image recognition to find specified object. The system can greatly increase security check efficiency with good image quality.