Can you share the face mask for the dataset

[07hyx06]

Hi, thanks for your excellent work!
I tried to obtain the face mask for each frame using the face-parsing.PyTorch repo as suggested in the paper, but the results are not as good as Fig 6, see the Obama example predicted directly from face-parsing.PyTorch:

Can you release the face mask or provide some insights on obtaining more accurate masks?

Thanks.

[XuanGhahahaha]

Hi, our segmented results are not more accurate than yours. Athough there will be a little noise and flicking in segmented GT face images, NeRF tends to synthesize a denoised and smooth result when training is finished. Just using these segmented images to supervise training is enough.

[07hyx06]

Got it, thanks for your reply