Description

This project aims to leverage a Coral TPU connected via USB to perform face recognition on a NAS running Nextcloud. Due to hardware limitations (ARM CPU with 1GB of RAM), standard face recognition plugins are not feasible, as they require a GPU and at least 2GB of RAM. Instead, the goal is to use TensorFlow Lite on the Coral TPU for efficient processing.

Goals

• Implement face recognition using TensorFlow Lite on the Coral TPU.
• Integrate the solution with Nextcloud's API to automatically tag images stored on the NAS.
• Optimize performance for limited hardware without GPU support.

Resources

• Coral USB Accelerator
• TensorFlow Lite documentation
• Nextcloud API reference
• Emoji Kitchen for a quick creative break when you need one!