What is the best facial recognition model from a library that any of you guys have worked with? It is a backup model that will be verifying matches coming from a lightweight model. I don’t mind if there is a high computational cost since the model will only be used to verify matches, I am worried more about accuracy. I am currently using SFace as the first model and VGG-Face through deepface as the second.
Here are some models which you can try
->FaceNet Keras: GitHub - nyoki-mtl/keras-facenet: Facenet implementation by Keras2
->David Sandberg’s FaceNet Implementation: GitHub - davidsandberg/facenet: Face recognition using Tensorflow
ArcFace & CosFace:
->Accessible through the InsightFace library: GitHub - deepinsight/insightface: State-of-the-art 2D and 3D Face Analysis Project
->VGGFace2 GitHub repository: GitHub - ox-vgg/vgg_face2