Method and system for estimating and tracking human body posture in video
A technology in human body posture and video, applied in the field of image processing and computer vision, can solve problems such as complex models, achieve the effect of improving accuracy, accurate estimation of human body posture, and inaccurate estimation of image blur
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0030] like figure 1 As shown, a human pose estimation and tracking method in a video, including:
[0031] Step (1) collects the video data in the human pose estimation data set;
[0032] Step (2) performing human body pose estimation on each frame image in the video data through the human body pose estimation network model to obtain estimated data, and obtaining optical flow estimation results between adjacent frames through an optical flow estimation algorithm;
[0033] Step (3) Input N frames of estimated data and the results of optical flow estimation between adjacent frames, a total of N+(N-1) parameters, into the improved CNN network model for human body pose tracking training, and perform overlapping based on loss function constraints Frame rectification outputs accurate human pose estimation and tracking data. where N≧2.
[0034] Further, it also includes preprocessing the video data, performing preprocessing on the video data to obtain a single frame image, and per...
Embodiment 2
[0049] A human body pose estimation and tracking system in video, comprising:
[0050] a data collection module, configured to collect video data in the human pose estimation dataset;
[0051] The data processing module is configured to perform human pose estimation for each frame of images in the video data through the human pose estimation network model to obtain estimation data, and obtain the optical flow estimation result between adjacent frames through an optical flow estimation algorithm;
[0052] The data correction module is configured to input N frames of estimated data and the optical flow estimation results between adjacent frames, a total of N+(N-1) parameters into the improved CNN network model for human pose tracking training, and based on the loss function Constrained overlapping frame correction outputs accurate human estimated pose and tracking data.
[0053] Further, the specific configurations of the data acquisition module, the data processing module and ...
Embodiment 3
[0055] A computer-readable storage medium for storing computer instructions, when the computer instructions are executed by a processor, the method for estimating and tracking a human body pose in a video as described in the above embodiments is completed.
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap