Skip to content

Latest commit

 

History

History
15 lines (10 loc) · 681 Bytes

README.md

File metadata and controls

15 lines (10 loc) · 681 Bytes

face_recognition

Some expirements on face recognition

5 celebrity dataset was taken from here

Face recognition can be devided in three steps:

  1. locate face
  2. ectract vector from face image
  3. compare vector with precomputed vectors

For the first step comparison of two face detection models faced and mtcnn were made.

Later I found a great library facenet-pytorch and used it for experiments.

Comparison of vectors were first made by cosine similarity and then with SVM model.