This is a fork of:

https://github.com/rwightman/posenet-python

Which is a port of:

https://github.com/tensorflow/tfjs-models/tree/master/posenet

Run:

python3 pose.py <path_to_image>

The programs will print found keypoins and write an output image into posenet.jpg

There is a flask web-service wrapping posenet estimation.

Run:

python3 app.py

It exports REST entry points:

• POST /pose/keypoints - get keypoints in JSON
• POST /pose/pic - get an image with drawn keypoints

There is a basic python client client/pose_client.py

Also could be called via curl:

• curl -X POST -F "file=@image.jpg" host:port/pose/keypoints
• curl -X POST -F "file=@image.jpg" host:port/pose/pic --output out.jpg

The flask service wrapped into a docker image.

• docker pull tprlab/tf-posenet-python
• docker run -p <host:port>:80 tprlab/tf-posenet-python