Hi, thanks for your great paper + code! It's very interesting way of solving the problem.
I want to try your method on my custom dataset, but I have one question about translation network (denoted as F in the paper).
It seems like there is no translation network used in BDL.py, and translated images are given in downloadable link.
If we want to apply custom dataset using this framework, how can we generate translated image using network F?
Is F trained separately and M is trained using translated image?
Hi, thanks for your great paper + code! It's very interesting way of solving the problem. I want to try your method on my custom dataset, but I have one question about translation network (denoted as F in the paper).
It seems like there is no translation network used in BDL.py, and translated images are given in downloadable link. If we want to apply custom dataset using this framework, how can we generate translated image using network F? Is F trained separately and M is trained using translated image?
Again, thanks for your interesting paper + code.