Thanks to Facebook research team who has created and open sourced Sapiens - Foundational human vision models. You can find their repository here: https://github.com/facebookresearch/sapiens.git.

This library is intended to create a wrapper library and offer Sapiens as API service.

Instruction to run

conda create -n sapiens_api python=3.10
conda activate sapiens_api
conda install pytorch torchvision torchaudio pytorch-cuda=12.1 -c pytorch -c nvidia

Install dependencies and download required model weight

pip install -r requirements.txt
python download_model.py

I am using the following version.

pytorch==2.5.1
torchvision==0.20.1
torchaudio==2.5.1

Running the API server via gunicorn

gunicorn main:app --workers 1 --worker-class uvicorn.workers.UvicornWorker --bind 0.0.0.0:8000 --daemon