search

ageitgey / face_recognition

The world's simplest facial recognition api for Python and the command line
MIT License
53.48k stars 13.49k forks source link

What is the principle of face comparison? Is the feature space distance? #654

Open LilyDreamZhao opened 6 years ago

LilyDreamZhao commented 6 years ago

What is the principle of face comparison? Is the feature space distance?

ageitgey commented 6 years ago

I'm not sure I totally understand the question, but my article here might have the answer you are looking for: https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78

The basic idea though is that the euclidean distance between two 128-feature face vectors should be less when the faces are more similar and larger when the faces are less similar. The model was trained to place faces of the same person within a euclidean distance of 0.6 of each other.

calmelo commented 3 years ago

I'm not sure I totally understand the question, but my article here might have the answer you are looking for: https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78

The basic idea though is that the euclidean distance between two 128-feature face vectors should be less when the faces are more similar and larger when the faces are less similar. The model was trained to place faces of the same person within a euclidean distance of 0.6 of each other.

I can't download the article, can you send the article to my mailbox. 269537259@qq.com