Human body motion posture and motion intention prediction method with infrared thermal imaging assistance

The invention discloses a human motion posture and motion intention prediction method with infrared thermal imaging assistance, and the method comprises the following steps: (1) converting infrared thermal imaging into a gray-scale map, multiplying the gray-scale map with groundtruth, and enabling a result to be called as softgroundtruth as a training label, (2) RGB images collected by softground and a camera respectively pass through an FCN network to take down and sample the last layer of output as image feature sequences, and the two obtained image feature sequences are subjected to attention so as to extract information related to a segmented image with infrared thermal imaging information from the RGB images, and (3) vectors obtained in the previous step are subjected to attention so as to obtain a segmented image with infrared thermal imaging information, and the segmented image with infrared thermal imaging information is extracted from the RGB images. And modeling is carried out on correlation between