Action recognition method and device, storage medium and electronic equipment
A technology of action recognition and recognition results, applied in the information field, can solve problems such as model overfitting, low accuracy of action recognition, uncontrollable built-in errors, etc., and achieve the effect of improving accuracy
Pending Publication Date: 2022-07-08
北京育达东方软件科技有限公司
View PDF0 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] However, in related technologies, the video action recognition technology based on human skeleton points cannot be controlled due to the built-in error of the human skeleton point detection model, and the error will be accumulated as noise and entered into the subsequent human action recognition model. Fewer and unbalanced samples may lead to overfitting of the model, resulting in lower accuracy of action recognition
Method used
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
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0062] The specific embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present disclosure, but not to limit the present disclosure.
[0063] figure 1 is a flowchart of an action recognition method according to an exemplary embodiment, and its execution subject may be a terminal, such as an electronic device such as a personal computer, a tablet computer, a mobile phone, or a server, which is not limited in the present disclosure. likefigure 1 As shown, the method includes the steps:
[0064] S101. Obtain a target video segment.
[0065] Wherein, the target video segment may be obtained by direct shooting through a mobile phone, a computer or a camera device, or may be obtained through network transmission of other electronic devices, etc., which is not limited in the present disclosure.
[0066] ...
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 PUM
Login to View More Abstract
The invention relates to an action recognition method and device, a storage medium and electronic equipment, and belongs to the technical field of information. Inputting the target video clip into a pre-trained skeleton point recognition model to obtain a skeleton point heat map corresponding to the target video; inputting the skeleton point heat map into a plurality of pre-trained action recognition models to obtain a plurality of recognition results, at least one of training methods, model structures and sampling frequencies of the action recognition models being different; and determining that the recognition result with the highest confidence degree in the plurality of recognition results is a target action type corresponding to the target video clip. The influence of overfitting is effectively overcome, and the accuracy of action recognition is improved.
Description
technical field [0001] The present disclosure relates to the field of information technology, and in particular, to an action recognition method, apparatus, storage medium, and electronic device. Background technique [0002] With the development of hardware computing power, human action recognition is becoming more and more popular in video classification, human-computer interaction and other fields. Among them, video action recognition based on human skeleton points has become a research hotspot, which is mainly attributed to the fact that skeleton points, as a compact representation of human actions, can effectively filter the noise in the original video. [0003] However, in the related art, the video action recognition technology based on human skeleton points cannot control the inherent error of the human skeleton point detection model, and the error will be accumulated into the subsequent human action recognition model as noise. The samples are small and unbalanced, ...
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 Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06V40/20G06K9/62G06N3/04G06N3/08G06V10/774G06V10/82
CPCG06N3/08G06N3/048G06F18/214
Inventor 陈奕名
Owner 北京育达东方软件科技有限公司