Introduction

degirum-face is a high-performance Python library for face detection, recognition, and tracking in images and video. Built for production deployments with minimal code and support for CPU and edge AI accelerators.

Fast and efficient face recognition with:

• Face detection, alignment, embedding extraction, and matching for images, video files, and live streams

• Multi-hardware support: CPU, GPU, and edge AI accelerators (Hailo, Axelera, DEEPX, Intel, NVIDIA, Rockchip, Google, DeGirum)

• Simple APIs: Minimal code to detect, enroll, and recognize faces with easy batch processing

• Flexible configuration: Python or YAML config for models, thresholds, and database paths

• Production-ready tracking: Real-time face re-identification, event notifications, and automated alert recording

• Robust database: LanceDB-based storage with vector similarity search

Start with Installation & Setup and Basic Concepts, then explore the Guides for your use case.