This project was made as part of my learning experince in trying to move my knowledge into 3D space when it comes to Computer Vision. It is seperated into 2 phases, the first one is successfully getting the pose estimation into the 3D space, the second phase will be getting the movements properly mapped to a 3D model.
- Connect the 2 cameras.
- Run
calibration_images
to capture the needed images.
- Press
s
to capture the images. - Run
cammera_calibration
to calibrate the cameras and save the file needed to calculate the distance of each keypoint. - Run
main
to start the process of 3D pose estimation.
- Connect the program with a 3rd party 3D program. DONE
- Visualize the movements with a 3D rigged model.
I'll soon explain the code and the algorithms used.
This will be the subject of major polish to make it more accurate, faster, and precise. Use as as a proof of concept