Setup

Create venv

Go to the root directory, and

python -m venv project_env

venv requirements

pip install -r requirements.txt

Note that the numpy version is not the latest

facebook repos

if not present: git clone detectron and videopose3d repos

git clone https://github.com/facebookresearch/VideoPose3D.git
git clone https://github.com/facebookresearch/detectron2.git

Detectron2 requirements

cd detectron2
pip install -e .

This will also setup the configs for the models. Then, go back to the root directory

Checkpoint for Videopose3d

Download the pretrained model:

bash download_checkpoint.sh

Current

If you stick to the main.py file, create a folder named current, with 2 subfolder input and output.

All good.

See the end of the main.pyfile to change the path