search

davidsandberg / facenet

Face recognition using Tensorflow
MIT License
13.81k stars 4.81k forks source link

Identification using a video stream #932

Open leetwito opened 5 years ago

leetwito commented 5 years ago

Hi,

I want to create a system for face identification, but use few frames for both acquisition and test time, in order to reduce errors.

  1. I couldn't find any work for comparing 2 populations instead of 2 points. Moreover, I would like to apply a threshold in order to add an 'Unknown' class. I would appreciate if you share any resources regarding those points.
  2. Can I assume that the embeddings of the same person would be Normally distributed? If so, why?
  3. Are there weights trained on grayscale images? or would it be good enough to duplicate the singe channel 3 times?

Thanks, Lee

RUDRANSH-hub commented 3 years ago

hi myself Rudransh Srivastava currently in 2nd year doing B-Tech with Information Technology branch. i had done car detection project under my IOT proff. i have idea about haar cascade frontal face detection and done a face detection project in deep learing by python. i wanted to serve my knowledge by contributing in your project. and fix it please let me know how should i contribute in this project.and give me chance to contrib.

thankyou