g = L2CS-Net / gaze estimation and tracking

[johndpope]

We extract g for the training video clips using [ 2] on each frame followed by a simple histogram-based clustering algorithm

https://github.com/Ahmednull/L2CS-Net
https://arxiv.org/abs/2203.03339

[johndpope]

fixed
Ahmednull/L2CS-Net#39