nntoolbox.vision.components.layers module

class nntoolbox.vision.components.layers.BiasLayer2D(out_channels: int, init: float = 0.0)

Bases: torch.nn.modules.module.Module

Add a trainable bias vector to input:

y = x + bias

forward(input: torch.Tensor) → torch.Tensor

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

training: bool

class nntoolbox.vision.components.layers.ConvolutionalLayer(in_channels, out_channels, kernel_size=3, stride=1, padding=0, bias=False, activation=<class 'torch.nn.modules.activation.ReLU'>, normalization=<class 'torch.nn.modules.batchnorm.BatchNorm2d'>)

Bases: torch.nn.modules.container.Sequential

Simple convolutional layer: input -> conv2d -> activation -> norm 2d

training: bool

class nntoolbox.vision.components.layers.CoordConv2D(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode='zeros')

Bases: torch.nn.modules.conv.Conv2d

Implement CoordConv https://arxiv.org/pdf/1807.03247.pdf

static augment_input(input)

Add two coordinate channels to input :param input: (N, C, H, W) :return: (N, C + 2, H, W)

bias: Optional[torch.Tensor]

dilation: Tuple[int, …]

forward(input)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

groups: int

kernel_size: Tuple[int, …]

out_channels: int

output_padding: Tuple[int, …]

padding: Tuple[int, …]

padding_mode: str

stride: Tuple[int, …]

transposed: bool

weight: torch.Tensor

class nntoolbox.vision.components.layers.CoordConvolutionalLayer(in_channels, out_channels, kernel_size=3, stride=1, padding=0, bias=False, activation=<class 'torch.nn.modules.activation.ReLU'>)

Bases: torch.nn.modules.container.Sequential

Simple convolutional layer: input -> conv2d -> activation -> batch norm 2d

training: bool

class nntoolbox.vision.components.layers.Flatten

Bases: torch.nn.modules.module.Module

forward(input)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

training: bool

class nntoolbox.vision.components.layers.HighwayConvolutionalLayer(in_channels, main)

Bases: nntoolbox.components.components.HighwayLayer

Highway layer (for images):

y = T(x) * H(x) + (1 - T(x)) * x

training: bool

class nntoolbox.vision.components.layers.InputNormalization(mean, std)

Bases: torch.nn.modules.module.Module

Normalize input before feed into a network Adapt from https://pytorch.org/tutorials/advanced/neural_style_tutorial.html

forward(img)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

training: bool

class nntoolbox.vision.components.layers.Reshape

Bases: torch.nn.modules.module.Module

forward(input, new_shape)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

training: bool