Viet-Anh on Software Logo

What is: Spatially-Adaptive Normalization?

SourceSemantic Image Synthesis with Spatially-Adaptive Normalization
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

SPADE, or Spatially-Adaptive Normalization is a conditional normalization method for semantic image synthesis. Similar to Batch Normalization, the activation is normalized in the channel-wise manner and then modulated with learned scale and bias. In the SPADE, the mask is first projected onto an embedding space and then convolved to produce the modulation parameters γ\gamma and β.\beta . Unlike prior conditional normalization methods, γ\gamma and β\mathbf{\beta} are not vectors, but tensors with spatial dimensions. The produced γ\gamma and β\mathbf{\beta} are multiplied and added to the normalized activation element-wise.