Facenet embeddings

I wanted to know

  1. What does facenet embeddings represent for an image containing multiple faces.
  2. Is is intuitive to use facenet embeddings for task other than face recognition for example emotion detector.

When I used facenet I did face detection, then face alignment then I passed this aligned face to facenet and that’s the way I have seen people use facenet, so there won’t be many faces in the input to facenet. For emotion detection also we can try doing this same thing detect faces, align them then do emotion detection. I think you can try fine tuning facenet to do emotion detection.

I want to know the concept of how to get the embedding values from given image in facenet?

1 Like