0
$\begingroup$

I am trying to make a face login application that authenticates the user when matching the registered face and the given face. currently, the issue is I cant extract the face descriptions from the given face when the user is taking the photo in the night or the photo has a backlight.

currently, I am using JavaScript API for face detection and face recognition in the browser and nodejs with tensorflow.js

can anyone suggest any good face detection and comparison algorithm that resolve my current issues, that will be very helpful for me

Now I am extracting face descriptions from the face and using the Euclidean distance equation is used for comparing the similarity of the images. if any good methods for comparison, please suggest

$\endgroup$
0
$\begingroup$

You can find a number of methods used for face detection. The simplest and most efficient way is to use embedding vectors of faces ( with FaceNet ).

  • Using the FaceNet model, we transform every face to a fixed dimensional vector. As the FaceNet model is trained on a large dataset, it's accuracy is trustworthy.

  • Now, we have two options. First, compare the vectors directly using cosine similarity or any such metric. Second, train a Siamese Model to differentiate between faces ( vectors )

Tip:

Siamese models would only work if you are authenticating a finite number of people. In order to add a new user, you need to retrain the whole model.

|improve this answer|||||
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.