Thanks for your interesting work, i am a little confused with this paragraph of codes, is that means each class in the semantic label map needs a linear layer? So, are there 150 linear layers in the ADE20k dataset??? Can you tell me how to implement your method in the ADE20k in details?
Thanks for your interesting work, i am a little confused with this paragraph of codes, is that means each class in the semantic label map needs a linear layer? So, are there 150 linear layers in the ADE20k dataset??? Can you tell me how to implement your method in the ADE20k in details?
https://github.com/ZPdesu/SEAN/blob/04c7536ff3fecd2d1a09c9ae046a1144636033a5/models/networks/normalization.py#L209