Face Recognizer

What is FaceRecognizer?

FaceRecognizer processes images and video frames independently without temporal tracking. It detects and identifies faces in single images, image batches, or video streams (frame-by-frame via predict_batch()).

When to Use FaceRecognizer

Choose FaceRecognizer when you need to:

• Process photo albums or image collections

• Analyze video files frame-by-frame without temporal tracking

• Run batch recognition on multiple images or video streams

• Build a face database from images

• Work without real-time tracking requirements

For video with persistent face tracking, use FaceTracker instead - it maintains face identities across frames and supports real-time alerts.

Core Concepts

Configuration

All settings are controlled through FaceRecognizerConfig:

import degirum_face

config = degirum_face.FaceRecognizerConfig(
    face_detection_model_spec=detector_spec,   # Detection model
    face_embedding_model_spec=embedding_spec,  # Embedding model
    db_path="./face_db.lance",                 # Database location
    cosine_similarity_threshold=0.6,           # Match threshold
)

face_recognizer = degirum_face.FaceRecognizer(config)

See Configuration Guide for all options.

Methods

FaceRecognizer provides methods for enrollment, prediction, and database management:

• enroll_image() / enroll_batch() - Add faces to database

• predict() / predict_batch() - Recognize faces in images/video

See Methods Reference for complete API documentation with examples.