Fig. 8
From: Real-world defocus deblurring via score-based diffusion models

Structure diagram of NCSN network. R, C and A represent residual block, convolutional layer, and attention block, respectively.
From: Real-world defocus deblurring via score-based diffusion models

Structure diagram of NCSN network. R, C and A represent residual block, convolutional layer, and attention block, respectively.