Method and system for detecting key points of human skeleton

Abstract

Embodiments of the present invention provide a method and system for detecting key points of human bones. Wherein, the method includes: inputting the human body image to be tested into the human skeleton key point detection model after training, and determining the human body bone key point of the human body image to be tested according to the output result; wherein, the human body bone key point detection model after training is composed of multiple The network structure of the asymmetric neural network at all levels in the multi-level asymmetric neural network is the same, and they all use dilated convolution to replace the convolutional layer and the pooling layer. In the method and system provided by the embodiment of the present invention, by inputting the image of the human body to be tested into the model, and determining the key points of the human skeleton according to the output result of the model, the model adopts an asymmetric multi-level repeated network structure, re-encodes and lightly decodes, and can Benefit more from the advantages of transfer learning, use the method of reducing the number of channels to improve the detection accuracy without affecting the speed.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of human skeleton key point detection, and in particular, to a method and system for detecting human skeleton key points. Background technique [0002] The key points of the human skeleton refer to the positions of the fixed parts of the human body, including eyes, nose, shoulders, waist, etc. The key points of the human skeleton are very important for describing the posture of the human body and predicting the behavior of the human body. Basics, such as action classification, abnormal behavior detection, automatic driving, etc. [0003] In recent years, with the development of deep learning technology, the detection effect of key points of human bones has been continuously improved, and it has been widely used in related fields of computer vision. The key point detection methods of human bones based on deep learning are mainly divided into two categories: top-down detection methods an...

AI Technical Summary

Problems solved by technology

[0004] For the single human skeleton key point detection in the top-down detection method, due to the weak distinction of the local information of the key points, the difficulty of detecting different key points is not the same, and the key point positioning of the human skeleton is more dependent on algorithm, therefore, there will be problems such as inaccurate detection and repeated detection

Images