Human body behavior recognition method based on deep learning

A deep learning and recognition method technology, applied in the field of deep learning, can solve problems such as insufficient training data, high computing and storage overhead, and difficulty in dynamic model update, achieving good recognition results, optimized speed, and improved accuracy
CN114220176APending Publication Date: 2022-03-22NANJING HOWSO TECH
0 Cites 11 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANJING HOWSO TECH
Publication Date
2022-03-22

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to a human behavior recognition method based on deep learning, and the method comprises the steps: S1, collecting data: collecting the data, and forming a data set; s2, data set processing: inputting data sets, performing personnel detection and tracking on data in the data sets, extracting skeleton information of each data set through human body posture estimation, and performing posture estimation to obtain a posture estimation result; and S3, dangerous behavior analysis: training and constructing an ST-GCN recognition model by using the data set, inputting the attitude estimation result into the ST-GCN recognition model for dangerous behavior analysis and recognition, obtaining a recognition result, and outputting the recognition result. According to the method, human body detection is carried out by utilizing a YOLO V4 algorithm of target detection, target tracking is carried out on a human body, a DeepSort tracking algorithm is adopted, bone joint points are extracted by utilizing OpenPose, and finally human body behavior recognition is carried out on a bone sequence by utilizing an ST-GCN recognition model.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention relates to the technical field of deep learning, in particular to a recognition method of human behavior based on deep learning. Background technique

[0002] Events caused by social security, school safety, road traffic, natural or man-made disasters, etc. have brought huge loss of life and property to people. The detection, alarm and processing of events have become particularly important. If only relying on manpower to complete these tasks, It can no longer meet people's needs, which requires technical means to assist people in completing incident detection, alarm and other related work, reducing people's work burden, and improving the efficiency of incident handling at the same time. At the same time, government departments also attach great importance to such incidents and require the use of "human defense, physical defense and technical defense" to efficiently deal with or even avoid such incidents.

[0003] With the development of...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More