Code Explanined??

[SalsHacker]

Hi David,I was wondering if you could explain the code to me and how the training works.
Thanks

[vudung45]

There is no actual training is done.

If you want to learn more about transfer learning, you can read up about it and use the output that Facenet model provides to transfer learning

[skt7]

Here no training is done. Actually, they have used a pre-trained FaceNet model which is designed to learn how to distinguish between two faces in Euclidean Space. The main idea behind this is Siamese Network and Triplet Loss

So now when you input a face image it produces a 128D embedding(vector) that represents that particular face. You can repeat it same for the other images and find the Euclidean distance between them it will give you the closeness score.