A Human Action Classification Method Based on Pose Recognition

Abstract

The invention discloses a human body action classification method based on gesture recognition, which includes the following steps: step 1, perform gesture recognition on the upper body of the human body, and obtain skeleton features that can represent the position, direction and size of each part of the upper body of the human body; step 2, Normalize the data in the skeleton features obtained in step 1; step 3, use multi-classification SVM to train the normalized skeleton features, and obtain a classifier that can classify different actions; step 4, use The classifier trained in step 3 classifies the input action. The collected human motion pictures are used as test data for experiments, and the experimental results show that the classification accuracy rate of the present invention reaches 97.78%, which can classify human motions well.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a gesture recognition-based human action classification method. Background technique [0002] The rapid development of computer network technology and multimedia technology has created convenient conditions for the storage and transmission of massive visual information such as images, and people can obtain a large amount of picture information from the Internet. However, the increasing amount of data also makes it difficult for people to find the pictures they want. For the website, it is necessary to manage this large amount of picture information, classify the pictures, and build an index, so that users can easily obtain the required content. For the majority of users, it is also hoped that they can find the picture information they need quickly and effectively, so as to reduce unnecessary waste of time. Therefore, classifying pictures has important practical signifi...

AI Technical Summary

Problems solved by technology

In terms of feature expression based on visual capture technology, human body contours were initially used as pose feature expressions. However, contour features describe poses from an overall perspective, ignoring the details of each part of the body, and cannot accurately represent rich and colorful human poses.

Images