The general process of thermal infrared human behavior recognition is to establish a reasonable behavior model based on the use of thermal imaging cameras to collect behavior sequence images, combined with thermal infrared imaging characteristics and human behavior characteristics, and then extract the behavior sequence images from the behavior sequence images. Describe the behavioral characteristics of the ability, and then use appropriate methods to train or match these characteristics, and then realize the recognition of human behavior types. From the perspective of information processing, this process can be divided into two parts: "behavior representation" and "behavior classification".