Request for comments on Face Detection API.

We presented the FaceDetection API [Link<https://github.com/w3c/mediacapture-image/issues/289>] sometime back along with a few other webrtc-NV features in the Breakout session at TPAC 2021 (Slides<https://docs.google.com/presentation/d/1aTxCqQkPoR9KajOs1PtBdjuW7ZvMDxAMLHyybPfZYsk/edit?usp=sharing> and minutes<https://www.w3.org/2021/10/20-webrtc-ic-minutes.html>).
A few comments regarding Face Detection API were -

  1.  Can we make the Face Detection a bit more general, say return a general shape instead of a bounding box ?
  2.  Can we get a mask, to be used as a stencil later for Background Blur, or a FaceMesh ?
  3.  Do we need to make the API a bit for forward looking instead of fixating on the present Platform APIs.

Tomorrow in the Nov 24 virtual meeting, we would like to solicit some feedback on the reworked API proposal [Link<https://github.eero.häkkinen.fi/intel-w3c-mediacapture-extensions/index.html#face-detection>]. We have tried to make it quite general, given the constraints of the present and very near term (upcoming) Platform APIs we are aware of. For e.g. FaceMesh is quite unlikely to be supported by Platform APIs anytime soon, but of course the API should be extendable to accommodate newer features.

I was planning to get a pulse check from the WebRTC community before putting up the PR where we can work on the finer details.

Best,
Riju.