About network architecture

[janchen0611]

Hi,
Congratulated for your team get the champion of AIM 2020 Image Inpainting Challenge.:)
Your result is really good and I would like to reproduce it.
After read your paper, I have some questions about network architecture:

1. How many DMFB do you use in the middle of network?

2. About the DMFB, the paper described: "Instance normalization (IN) and ReLU activation layers followed by the first convolution, second column convolutions and concatenation layer are omitted for brevity. The last convolutional layer only connects an IN layer."
   Does it means the first conv-3 is "Conv-IN-Relu" and the last conv-1 is "Conv-IN"?

3. Continued by Q2, the last Add is before or after the IN?

Thanks for your help.

[Zheng222]

@janchen0611 Thank you for your interest in our method.

1. 8 DMFBs in our experiment
2. Yes, your understanding is absolutely correct.
3. after IN