Infrared optical imaging payloads make use of the thermal radiation intensity characteristics of the target for imaging,which has a certain degree of falsification capability and can circumvent the limitations of visible light imaging equipment that cannot be imaged at night and under adverse meteorological conditions.However,the infrared image has low contrast and unclear edges,which greatly reduces the accuracy of imaging target recognition.In this paper,an infrared vehicle target detection method is proposed based on YOLOv5.Firstly,the RFBs module in the shallow feature layer is introduced to improve the receptive field and detection effect of small targets.Secondly,the BiFPN structure is used to realize the re use of the underlying features in order to fuse more features,and the CBAM attention mechanism is added to improve the detection accuracy in the neck network(Neck)part.The experimental results show that the detection effect on the Drone Vehicle dataset is better than that of the original network,the precision rate(Precision)is improved by 2.8%,the recall rate(Recall)is increased by 16%,and the average precision(mAP)is increased by 2.3%.It is concluded that it can be effectively applied to the autonomous detection and recognition of vehicles in aerial infrared images.