Markless Motion Analyser

Version 6.0.0 (37,5 ko) par Rodrigo Bini
This program extracts keypoints (i.e. joints) using a pre-trained neural network and calculates body CoM location.
0.0
(0)
97 téléchargements
Mise à jour 15 août 2022

Afficher la licence

This app expands from an existing code (https://github.com/matlab-deep-learning/Human-Pose-Estimation-with-Deep-Learning) to extract keypoints using a pre-trained neural network and calculates body CoM location from video files. It does the same by integrating a module provided in this link: MATLAB - OpenPose - wrapper - File Exchange - MATLAB Central (mathworks.com)
There is a module to create videos with the joints and centre of mass locations and another module to scale the coordinates using linear calibration and scale the coordinates using a checkerboad calibration.
To use the OpenPose module, you need to download files provided by Gil: GitHub - gilserrancoli/capture_2Dcycling: This repository contains the MEX function to call OpenPose routine for image processing, and the MATLAB code to analyze the cycling data

Citation pour cette source

Rodrigo Bini (2024). Markless Motion Analyser (https://www.mathworks.com/matlabcentral/fileexchange/101784-markless-motion-analyser), MATLAB Central File Exchange. Récupéré le .

Compatibilité avec les versions de MATLAB
Créé avec R2022a
Compatible avec les versions R2019b et ultérieures
Plateformes compatibles
Windows macOS Linux

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!
Version Publié le Notes de version
6.0.0

Addition of module to run OpenPose and to scale coordinates using checkerboard.
5.0.0

Amended image resize. Needs to export images for scaling after passing them through the network to match image resolution. For better results, select the whole screen for cutting the image.
4.0.0

Amending reference to source code.
3.0.0

Corrected image resolution after CNN tracking.
2.0.0

Corrected image resolution in the CNN output.
1.0.0