spandrel.architectures.__arch_helpers.block.ResidualDenseBlock_5C

class documentation

class ResidualDenseBlock_5C(nn.Module): (source)

Constructor: ResidualDenseBlock_5C(nf, kernel_size, gc, stride, ...)

View In Hierarchy

Residual Dense Block Style: 5 convs The core module of paper: (Residual Dense Network for Image Super-Resolution, CVPR 18) Modified options that can be used:

  • "Partial Convolution based Padding" arXiv:1811.11718
  • "Spectral normalization" arXiv:1802.05957
  • "ICASSP 2020 - ESRGAN+ : Further Improving ESRGAN" N. C. {Rakotonirina} and A. {Rasoanaivo}
Parameters
nfChannel number of intermediate features (num_feat).
gcChannels for each growth (num_grow_ch: growth channel, i.e. intermediate channels).
convtypethe type of convolution to use. Default: 'Conv2D'
gaussian_noiseenable the ESRGAN+ gaussian noise (no new trainable parameters)
plusenable the additional residual paths from ESRGAN+ (adds trainable parameters)
Method __init__ Undocumented
Method forward Undocumented
Instance Variable conv1 Undocumented
Instance Variable conv1x1 Undocumented
Instance Variable conv2 Undocumented
Instance Variable conv3 Undocumented
Instance Variable conv4 Undocumented
Instance Variable conv5 Undocumented
def __init__(self, nf=64, kernel_size=3, gc=32, stride=1, bias: bool = True, pad_type='zero', norm_type=None, act_type='leakyrelu', mode: ConvMode = 'CNA', plus=False): (source)

Undocumented

def forward(self, x): (source)

Undocumented

conv1 = (source)

Undocumented

conv1x1 = (source)

Undocumented

conv2 = (source)

Undocumented

conv3 = (source)

Undocumented

conv4 = (source)

Undocumented

conv5 = (source)

Undocumented