v0.5 - The bloat update

Hey bro, I heard you like bloat!
Here's some bloat!

Compressed: 2.78GB
Uncompressed: 5GB 💀

Because of Tensorflow and Onnx, the whole CUDA toolkit has to be included.
I don't know the legality of this, but this should include the whole cuda toolkit in runtime form.
The sources for the .dll is from nvidia's official PYPI

Download

Github has a file size limit of 2GB.
Download here

(.exe not working) v0.4 - GPU Support

Since this is using ONNX, this should work with any popular GPU. Examples:

• Nvidia: Cuda
• AMD: ROCm
• Intel
  Full list should be here

Apparently, my last version didn't include GPU support and therefore only was 0.66GB

(.exe not working) v0.3

Spent multiple hours completely test the program.
i'm not quite sure how, but it is somehow only 0.67GB now. (was 1.7GB in v0.2)
This version has been tested on my Main, Laptop and frash VM

NOTE: I should have included a loading text. It is loading, just really slow because of compression. I'm adding GPU support in a day or two

(.exe not working) v0.2

• Slightly faster infrence
• Infrence limitor setting
• UVC support
• Serial camera support
• Slightly more orginized

(.exe not working) v0.1

No UI yet, come back later.
This supports CPU and CUDA.
Change settings in Settings.json

Ryan's eyetracking uses ~13% CPU and ~23% GPU
My version uses ~1.5% CPU and ~5% GPU

My version

Ryan's vesion