Multi-person human body posture estimation method

Abstract

The invention discloses a multi-person human body posture estimation method, and relates to processing of a recording carrier for identifying graphics. The method is a multi-person human body postureestimation method based on depth features. According to the method, a depth feature human body key point detection model composed of a main body network and a fine tuning network is constructed; a depth feature human body key point detection model is trained by adopting a two-stage training method combining a bottom-up method and a top-down method; finally, the human body key points are detectedby using a depth feature human body key point detection model trained in two stages. Redundant key points which do not belong to a target person are removed through human body key point clustering processing, then a multi-person human body posture estimation result is output, and the defects that in the existing multi-person human body posture estimation method technology, under the condition thatcrowds are dense, the target human body shielding interference robustness is poor, and the human body key point detection accuracy is low are overcome.

Description

technical field [0001] The technical solution of the present invention relates to the processing of a record carrier for recognizing graphics, in particular to a method for estimating the pose of a multi-person human body. Background technique [0002] Human body posture estimation technology is very important to describe the trajectory of human motion and predict human behavior. This technology is used to locate the motion trajectory of key points describing the position of human joints and record its motion data, realize 3D animation to simulate human motion to make movies and TV, and pass The recorded human motion trajectory and data are used to analyze human motion, which is applied to the classification of human motion, the detection of abnormal human behavior, and the field of automatic driving. [0003] The existing human pose estimation methods are divided into machine learning methods and deep learning methods according to the different feature definition methods. ...

AI Technical Summary

Problems solved by technology

[0016] The technical problem to be solved by the present invention is to provide a method for estimating the pose of a multi-person human body, which is a method for estimating the pose of a multi-person human body based on depth features. The key point detection model uses a two-stage training method combining bottom-up and top-down methods to train the deep feature human key point detection model, and finally uses the two-stage trained deep feature human key point detection model to detect human key points. And the redundant key points that do not belong to the target person are removed through the clustering process of human body key points, and then the result of multi-person human body pose estimation is output, which overcomes the problem of the existing multi-person human body pose estimation technology in the case of dense crowds. Poor robustness to human body occlusion interference, and low accuracy rate of human body key point detection

Images