training from scratch

[liminn]

Thanks for great work~
I am trying to training your model from scratch(only use pretrained mobilenet weights).
I encountered two problems:

• The first problem is shown by the red arrow: the alpha value of the unkonwn region is not large enough.
• The second problem is shown by the blue arrow: outside the unknown region, there are always white scattered dots.

For the first problem, I think my training epochs is not enough (just trained to the 6th epoch).
For the second problem, I am very confused and have no ideas.
Do you have any suggestions on these two problems?

[poppinace]

Hi @liminn,

In my opinion, I think your ground truth alpha mattes are not generately correctly. Please check your code making sure the alpha is within the range [0, 1].

Hao

[liminn]

@poppinace Thanks, I will check the alpha labels.

[facetohard]

Hello @liminn @poppinace , I am trying to train indexnet from scratch. But I got worse performance than deep image matting by using the paper learning rate config。
So I wonder that how you trained indexnet.
Does you train with multi stage just like the paper "Deep Image Matting" does.