Overview

This repo contains simple code for using OpenPose pose detection inside a Jupyter Notebook. You will need to install Docker and Docker Compose before running this code.

Note on hardware: The docker images are built with the expectation of a CUDA enabled GPU being present on the host. Code was developed on a machine with an NVIDIA RTX 2080ti. The code might still work with CPU only, but I have not tested this.

Quickstart

From the project root dir, run:

# Note: Depending on your Docker Compose version, you may need to run 'docker-compose' instead of 'docker compose'
docker network create openpose_net
docker compose up   # Add the -d flag if you want to run in detached mode

Alternatively, if you want to make modifications and build the images from the local dockerfiles, use:

# Note: Depending on your Docker Compose version, you may need to run 'docker-compose' instead of 'docker compose'
docker compose -f docker-compose.build.yaml up --build   # Add the -d flag if you want to run in detached mode

You should now be able to visit http://localhost:8888/lab in your browser and find the Jupyter server running there. The default password for the server is openpose. You can change this by following the instructions here, updating the jupyter_notebook_config.json file with your hashed password, and rebuilding the image.

To stop the running containers:

docker compose down