yihuacheng / dvgaze Goto Github PK

Thank you for the interesting work!

I have one question regarding the following part in Section 5.1 of your paper:

Therefore, we re-calculate gaze directions in the two datasets. More concretely, given a pair of images, we get the corresponding 2D facial landmarks from datasets and estimate 3D landmarks using triangulation. We select nose as gaze origin since nose is usually visible in all views. We compute new gaze directions with the new gaze origin.

This recalculation is done using the raw ETH-XGaze dataset, not the preprocessed dataset (which does not include landmark data), correct? Does it also involve re-rectifying the images into cropped 224x224 face patches based on the new triangulated landmarks?
Is the preprocessing code to perform this calculation available somewhere? I did not find it in this repo nor on the Gazehub website.

请问可以用你这边开源的模型测试我自己的视频吗？还是说由于在数据预处理那一块的原因呢，Gaze360数据集上训练的模型就仅仅可以在Gaze360上做测试呢？其他的数据集也同理

Hi, thank you for your perfect work~. When will you open source the code?

Thank you once again for your excellent work. When will you update the full version? I'm Looking forward to your next work.