Human body key point detection method and device, storage medium and terminal equipment

Abstract

The invention discloses a human body key point detection method. The method comprises the steps of obtaining a to-be-detected human body image; based on a preset key point detection network, obtaining a human body key point prediction heat map according to the to-be-detected human body image, wherein the key point detection network takes a preset Gaussian response heat map of a human body part as supervision information, and the Gaussian response heat map is pre-constructed according to a connection relationship between a human body data set and preset human body key points; and detecting the positions of the human body key points according to the human body key point prediction heat map. Correspondingly, the invention further discloses a human body key point detection device, a computer readable storage medium and terminal equipment. By adopting the technical scheme of the invention, human body key point detection can be carried out in combination with constraints between human body structure information, thereby reducing quantization errors and improving the accuracy of a detection result.

Description

Technical field[0001]The present invention relates to the field of image processing, and more particularly to a key point detection method, apparatus, computer readable storage medium, and a terminal device.Background technique[0002]The human key point detection is designed to detect the specific position of the articles, facial features, etc. of the image, to describe the human attitude, in the natural image, the complexity of shooting scenes and the human gesture diversity gives a huge challenge to the human attitude estimation task . As one of the most challenging topics in the field of computer visual technology, diversified human gesture estimates the emergence of data sets and the development of deep learning techniques, making the accuracy of human key detection are gradually improved, commonly used representative Human key detection network has HOURGLASS, CPN, MSPN, and HRNET, etc., these key points detection network based on confidence hot maps, mostly use multi-level featu...

AI Technical Summary

Problems solved by technology

[0002] Human key point detection aims to detect the specific positions of the joints and facial features of the human body on the image to describe the posture of the human body. In natural images, the complexity of the shooting scene and the diversity of human postures have brought great challenges to the task of human posture estimation.

As one of the most challenging topics in the field of computer vision technology, the emergence of diverse human pose estimation data sets and the development of deep learning technology have gradually improved the accuracy of human key point detection. The commonly used representative Human body key point detection networks include Hourglass, CPN, MSPN, and HRNet. These human body key point detection networks based on confidence heat map regression mostly use multi-level feature fusion modules to capture rich multi-scale information, but sacrifice the network. Complexity, ignoring the spatial constraints of human body structure information, resulting in certain quantitative errors when obtaining the specific positions of key points, making it difficult for human body key point detection technology to meet the needs of high accuracy in practical applications

Images