I made a change in the code, so now, if you want to load a single video, on the load button you choose a video.
If you want to load lots of videos, just choose a folder and the script will load all the mp4 files i that folder and create a mocap version of it, keeping each video file in a collection with the name of the video
The easiest way to install the python packages inside blender to be able to use this blend file and run your own pose estimation, is doing this:
go to your blender installation folder, and copy its path
In this example, the path to blender is:
D:\Blender\Blender_exe\blender-2.93.1-windows-x64\2.93
replace the path below with the one in your system:
D:\Blender\Blender_exe\blender-2.93.1-windows-x64\2.93\python\bin\python.exe D:\Blender\Blender_exe\blender-2.93.1-windows-x64\2.93\python\lib\site-packages\pip install mediapipe
D:\Blender\Blender_exe\blender-2.93.1-windows-x64\2.93\python\bin\python.exe D:\Blender\Blender_exe\blender-2.93.1-windows-x64\2.93\python\lib\site-packages\pip install opencv-python