Human body key point detection method based on attention residual module and branch fusion

Abstract

The invention discloses a human body key point detection method based on an attention residual module and branch fusion. The method belongs to the technical field of computer vision, and comprises thefollowing steps: performing feature processing on an input picture by using a feature extraction network to obtain a feature map; inputting the feature map into a region generation network to obtaina target suggestion box; performing region pooling operation to obtain a region-of-interest feature map; inputting the image into a convolution layer to carry out feature extraction operation to obtain a feature map I; performing feature extraction and fusion by using the branch I and the branch II; superposing the results of the two branches, carrying out resolution restoration through deconvolution, and then carrying out double linear interpolation up-sampling ; and modeling the positions of the key points into a one-hot binary mask for training. According to the invention, the diversity ofinformation output by the network is improved, different visual fields are better captured, the problem of disorder of detection key points is effectively solved in a simple scene, the accuracy and efficiency of the method are improved, and the method can well adapt to a complex scene.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a human body key point detection method based on an attention residual module and branch fusion. Background technique [0002] In the existing technology, the detection of human body posture has been widely concerned by domestic and foreign scholars, and it is also an important topic in the field of computer vision. In recent years, related research at home and abroad has divided the research methods of human body posture detection into wearable sensors and Based on the method in the field of computer vision; the former is mostly a contact posture analysis system, which has a high ability to analyze the human body, but the sensor for collecting human motion parameters needs to be attached to the human body or accessories, which is inconvenient to wear and brings people a lot of trouble. It feels unnatural; and its cost is high, difficult to operate, not suita...

AI Technical Summary

Problems solved by technology

[0003] In view of the above problems, the present invention provides the purpose of the present invention is to provide a human body key point detection method based on the attention residual module and branch fusion, which solves the problems of poor detection effect and low accuracy in the prior art. The problem

Images