Why do need to use the reflection padding, how does it help in the process? Is the relation completely empirical, or the idea is to use the image (input) relevant pixels than any other padding?
Thank you.
Hi Sujeet!
This is a line from the paper which states its purpose : We use reflection padding to reduce boundary artifacts
You can go through the paper for all the details : Pix2PixHD
Regards,
Nithin