I have got numerous frames and I've detected all the faces in all the frames using Retinaface. However I need to track the faces of people over frames.
For this purpose, I assumed I could try finding the landmarks from the face using libraries like dlib and maybe compare these landmarks to check if they are infact the face of the same person.
I would like to know if there are other methods or some useful resources I could refer for the same. Thanks a lot in advance.
The topic of your problem is person re-identification. You can check here.
You can try using something called as a Siamese network if you are willing to train the network on your own using something called as triplet loss(if you have lots of face images).
Another approach would we something called a one-shot using FaceNet(transfer learning approach) FaceNet uses deep convolutional neural network (CNN). The network is trained such that the squared L2 distance between the embeddings correspond to face similarity. The images used for training are scaled, transformed and are tightly cropped around the face area.
Another important aspect of FaceNet is its loss function . It is already trained using triplet loss function.In this case you could just feed to face images and you would get a thershold score for the similarities.
