The models and training code will be released.
We produce the code which making Fully Synthetic Document Shadow Removal Dataset (FSDSRD) for document shadow removal.
Direct link and description is in here.
We found wrong with our paper. Please check #5 (comment).