This repository is for a project detecting static human joint parameters in the gym. The basic skill used here is called human pose estimation/human skeleton tracking.
There are several ways to accomplish skeleton tracking.
All methods can mainly be divided into two types: RGBD-based and RGB-based.
- RGBD-based type means getting 3D human joint keypoints from the combination of a 2D image and a depth image.
- RGB-based type means getting 2D/3D human joint keypoints from a 2D image (with neural networks) and calculating the joint coordinates with depth data.
Here are some tutorials to implement the methods.
Hopefully I can get an EI conference paper for my master's degree with this project.