Method __init__

models/networks.py:341–400

Construct a Unet submodule with skip connections. Parameters: outer_nc (int) -- the number of filters in the outer conv layer inner_nc (int) -- the number of filters in the inner conv layer input_nc (int) -- the number of channels in input images/features

(self, outer_nc, inner_nc, input_nc=None, submodule=None, outermost=False, innermost=False,
                 norm_layer=RAIN, use_dropout=False, use_attention=False, enc=True, dec=True)

        |-- downsampling -- |submodule| -- upsampling --|
    """
    def __init__(self, outer_nc, inner_nc, input_nc=None, submodule=None, outermost=False, innermost=False,
                 norm_layer=RAIN, use_dropout=False, use_attention=False, enc=True, dec=True):
        """Construct a Unet submodule with skip connections.

        Parameters:
            outer_nc (int) -- the number of filters in the outer conv layer
            inner_nc (int) -- the number of filters in the inner conv layer
            input_nc (int) -- the number of channels in input images/features
            submodule (UnetBlockCodec) -- previously defined submodules
            outermost (bool)    -- if this module is the outermost module
            innermost (bool)    -- if this module is the innermost module
            norm_layer          -- normalization layer
            user_dropout (bool) -- if use dropout layers.
            enc (bool) -- if use give norm_layer in encoder part.
            dec (bool) -- if use give norm_layer in decoder part.
        """
        super(UnetBlockCodec, self).__init__()
        self.outermost = outermost
        self.innermost = innermost
        self.use_dropout = use_dropout
        self.use_attention = use_attention
        use_bias = False
        if input_nc is None:
            input_nc = outer_nc
        self.norm_namebuffer = ['RAIN', 'RAIN_Method_Learnable', 'RAIN_Method_BN']
        if outermost:
            self.down = nn.Conv2d(input_nc, inner_nc, kernel_size=4, stride=2, padding=1, bias=use_bias)
            self.submodule = submodule
            self.up = nn.Sequential(
                nn.ReLU(True),
                nn.ConvTranspose2d(inner_nc * 2, outer_nc, kernel_size=4, stride=2, padding=1),
                nn.Tanh()
            )
        elif innermost:
            self.up = nn.Sequential(
                nn.LeakyReLU(0.2, True),
                nn.Conv2d(input_nc, inner_nc, kernel_size=4, stride=2, padding=1, bias=use_bias),
                nn.ReLU(True),
                nn.ConvTranspose2d(inner_nc, outer_nc, kernel_size=4, stride=2, padding=1, bias=use_bias)
            )
            self.upnorm = norm_layer(outer_nc) if dec else get_norm_layer('instance')(outer_nc)
        else:
            self.down = nn.Sequential(
                nn.LeakyReLU(0.2, True),
                nn.Conv2d(input_nc, inner_nc, kernel_size=4, stride=2, padding=1, bias=use_bias),
            )
            self.downnorm = norm_layer(inner_nc) if enc else get_norm_layer('instance')(inner_nc)
            self.submodule = submodule
            self.up = nn.Sequential(
                nn.ReLU(True),
                nn.ConvTranspose2d(inner_nc * 2, outer_nc, kernel_size=4, stride=2, padding=1, bias=use_bias),
            )
            self.upnorm = norm_layer(outer_nc) if dec else get_norm_layer('instance')(outer_nc)
            if use_dropout:
                self.dropout = nn.Dropout(0.5)

        if use_attention:
            attention_conv = nn.Conv2d(outer_nc+input_nc, outer_nc+input_nc, kernel_size=1)