humanoidsctu/smplifyx repository overview

This projects contains part of our internal processing pipeline (smplify-x + SMIL) to extract infant pose estimation from videos, from which their motion can be derived, among other characteristics.

Note that while this repository is public out of convenience, it is primarily directed to our internal use can be modified at any time, without notice or description of the changes.

The image contains an extensive readme, but we will not publicly support it as the full pipeline also uses scripts that are described in the readme but not currently shared publicly, e.g. neither in this repository or in the internal_pipeline version of our vitpose docker image⁠.

This does not require a gpu, and the image can be started using the following command:
sudo docker run --rm -it -e="DISPLAY" -v=/tmp/.X11-unix:/tmp/.X11-unix:rw -v /home/$user:/home/$user humanoidsctu/smplifyx:internal_pipeline /bin/bash
Replace $user by the session's username.
Add -v /media:/media to give Docker access to external hard drives e.g. if that's where the data to process is.