a question about prepossessing
Closed this issue · 3 comments
hello hao,
Thanks for your third awesome work for document image dewarping. I have a simple question, that is,
In your section 5.2 of the paper , you pointed out that The preprocessing module and the following rectification module are trained independently
. as shown in the following:
My question is why don't you train the whole flow jointly? Have you made any trials to verify the 2-stage rectification would be better?
sorry, the issue title should be preprocessing, instead of prepossessing
Hi, thanks for your attention to our work, and this is a nice concern.
In fact, the whole network can be trained jointly.
We perform the 2-stage rectification here based on the following reasons,
- The gradients can be backpropagated to the prepossessing network due to the nondifferentiable mask. Because the distorted image should be multiplied by the document mask(0 denotes the background and 1 denotes the foreground document region).
- Joint training would consume more GPUs.
I hope this can help you.
I see, it helps me a lot, thanks for your kind response