Project Description
I have all my photos on a private NAS running nextcloud.
This NAS has an ARM CPU and 1GB of RAM, which means I cannot run the face recognition plugin because it requires a GPU, 2 GB of RAM, and PDLib is not available for this arch (I know I could build it and package it ... but doesn't sound fun ;) )
However, I have a Coral TPU connected to a USB port (Thanks to my super friend Marc!):
https://coral.ai/products/accelerator
Where I could run Tensorflow Lite... you see where this is going, don't you?
Goal for this Hackweek
The goal is to run face recognition on the Coral TPU using tensorflow lite and then using the nextcloud API to tag the images.
Resources
No Hackers yet
This project is part of:
Hack Week 20
Activity
Comments
-
16 days ago by zetisno | Reply
Users can Emoji Kitchen select two different emojis and blend them together to create a new character or avatar. For example, combining a heart emoji with a smiley face might result in a heart-eyed smiley, or blending a cat with a crying face could create a crying cat emoji.
-
2 days ago by Miranda23 | Reply
TPUs can greatly accelerate machine learning tasks, Geometry Dash Lite especially for inference tasks like face recognition. Google’s Coral USB Accelerator or Edge TPU provides hardware-accelerated inference with TensorFlow Lite models.
Similar Projects
This project is one of its kind!