Initializes a FaceRecognitionSystem class object.
-
Calls MTCNN face detection model object.
-
Calls FaceNet or VGGFace model to extract embeddings (features) from face images.
-
Creates a connection to the database of known faces by calling the "Database" class object.
- self.db is the dictionary of known faces with id : name key-value pairs.
- self.embeddings is the dictionary of known faces with id : embeddings key-value pairs.
- Loads face classifier that was trained on the database of known faces.
frsystem.frs.FaceRecognitionSystem(self,
embedding_model=None,
weights=None,
face_classifier=None,
**kwargs)| Name | Description |
|---|---|
| embedding_model | Options: 1. None. If you want to use only face location and facial features detection functionality. 2. facenet. Use FaceNet as the feature extractor model. Input size for FaceNet is 160x160x3 3. vggface. Use VGG-Face as the feature extractor model. Input size for VGG-Face is 224x224x3 |
| weights | File path to the weights for the chosen embedding model. Defaults to None |
| face_classifier | File path to pre-trained face classifier. Face classifier |
| **kwargs | Two keyword arguments that are passed to the Database class. db_file and embeddings_file |
More extended docs coming soon.
See https://github.com/amac-lfc/frsystem/tree/master/frsystem frs.py file for more information.