Abstract
The gait is a special feature that needs to be identified as it affects many parts of the body. Each person's gait is individual. Unlike fingerprint or retinal identifiers, gait can be recognized at a great distance without direct contact. Also, gait recognition during epidemic periods is more relevant than face recognition. Using machine learning algorithms, you can train a neural network to recognize the identity of each person. The study considers splitting the stream into frames and various options for background segmentation such as GrabCut and Mask R-CNN. What follows is a detailed analysis of the optimal solution methods for human personality recognition based on the outline and the bounding box.