This is a fork of Exadel-Inc CompreFace, a powerful tool with face recognition and face verification capabilities. Code are modified to accommodate CPU with more and better resources ( Xeon v3 , AVX2 ). Model and plugin in use: mobilenet 1.20 , retina_mnet_v1, arcface-r50-msfdrop75.

You can change the model or plugins according to your need. For more, please refer to here

• Low Resource Requirements: Fully capable of running on CPU with minimal hardware demands.
• Optimized for Dockge: The code is optimized to run on Dockge, which operates on top of Proxmox LXC.
• No GPU Required: Designed to work efficiently without the need for a GPU.

When integrated with Home Assistant and Frigate, the detector's inference speed is typically less than 100ms, ensuring fast and reliable performance.