I want to use YOLO to train a model for face recognition, aiming for the highest possible accuracy in distinguishing between different individuals, including recognizing faces from different angles and sides. Would you recommend training the model here on Roboflow? Or are there better alternatives for achieving this goal? And which YOLO version is best for this use case?
Probably better to use a purpose built facial recognition model, something like this GitHub - serengil/deepface: A Lightweight Face Recognition and Facial Attribute Analysis (Age, Gender, Emotion and Race) Library for Python
But if you want to use more off the shelf stuff like YOLO, you could do
- Face detection (train a YOLO on large dataset of faces so a model can detect any face)
- Feature extraction (use CLIP to get embeddings of the cropped face detection from step 1)
- Comparison (compare faces to get results of which face belongs to which person)