Architectures module

fc_lower_stack(depth, img_size_x, n_out_channels)	Returns simple architecture with single Fully Connected layer and Relu activatation.
osrci_lower_stack(depth, img_size_x, ...)	Returns architecture based on research in OSRCI method.
alexNet32_lower_stack(n_out_channels)	Returns modified architecture based on AlexNet, suitable for size (32 x 32 x 3) images.

torchosr.architectures.alexNet32_lower_stack(n_out_channels)

Returns modified architecture based on AlexNet, suitable for size (32 x 32 x 3) images.

Parameters:

n_out_channels (int) – Size output

Return type:

torch.nn.Sequential

Returns:

Lower stack sequential architecture

torchosr.architectures.fc_lower_stack(depth, img_size_x, n_out_channels)

Returns simple architecture with single Fully Connected layer and Relu activatation.

Parameters:

• depth (int) – Number of color channels

• img_size_x (int) – Size of image in single axis

• n_out_channels (int) – Size output

Return type:

torch.nn.Sequential

Returns:

Lower stack sequential architecture

torchosr.architectures.osrci_lower_stack(depth, img_size_x, n_out_channels)

Returns architecture based on research in OSRCI method.

Parameters:

• depth (int) – Number of color channels

• img_size_x (int) – Size of image in single axis

• n_out_channels (int) – Size output

Return type:

torch.nn.Sequential

Returns:

Lower stack sequential architecture