This repository contains code to use Microsoft Azure's cognitive services face API to detect faces, facial landmarks etc, and output them in a format suitable for abc-display-tool.
To get started, edit the writeapi.py file to include your API key. Run this to generate an api.p "pickle" file. This should not be uploaded to git.
Usage:
msface.py [-h] --videofile VIDEOFILE [--outfile OUTFILE] --startframe
STARTFRAME --endframe ENDFRAME [--APIpickle APIPICKLE]
A csv file will be output to OUTFILE containing the facial features
detected. One row will be output per face per frame
TODO: decide whether to handle multiple faces per frame here, or further down the pipeline