how to do some test like Fig 8 in your paper

[Johnson-yue]

Hi， thanks your sharing ，I want to test about Digital watermarking

Would you provide some tutorial for this part in readme or somewhere？ thank you

[jiupinjia]

Hi @Johnson-yue, thanks for your feedback and sorry for the late reply. Currently, we haven't released any of our pre-trained de-watermarking models. But you can still train one from scratch. To do this, you need to first download our experimental dataset and then run the following command:
python train_decoder.py --dataset watermarking --net_G unet_256 --in_size 256 --batch_size 16 --norm_type batch --with_disparity_conv --checkpoint_dir ./checkpoints_your_model_name_here --vis_dir ./val_out_your_model_name_here
Please refer to this tag in the readme file for more details.
I will be releasing all of our pretrained models in our paper, hopefully, in the near future.