cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers

A modified version of the Google UVQ source file.

As the original file is under apache lisence, I should mention that this is a modified version of the source file:

https://github.com/google/uvq/blob/main/uvq_pytorch/utils/custom_nn_layers.py

Classes

Conv2dNormActivationSamePadding(in_channels, ...)	Part of uvq model.
Conv2dSamePadding(in_channels, out_channels, ...)	Part of uvq model.
Conv3DSamePadding(in_channels, out_channels, ...)	Part of uvq model.
InceptionB0B3(conv1_in, conv1_out, pool_shape)	Part of uvq model.
InceptionB1B2(conv1_in, conv1_out, ...)	Part of uvq model.
InceptionMixed(conv1_in, b0_conv1_out, ...)	Part of uvq model.
InceptionMixedBlock()	Part of uvq model.
Interpolate([size, scale_factor, mode, ...])	Imitate the BilinearResize layer of tensorflow.
MBConvSamePadding(input_channels, ...)	Part of uvq model.
MaxPool3dSame(kernel_size, stride)	Part of uvq model.
PermuteLayerNHWC([permutation])	Part of uvq model.

Details

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.Conv2dNormActivationSamePadding(in_channels: int, out_channels: int, kernel_size: int | ~typing.Tuple[int, int] = 3, stride: int | ~typing.Tuple[int, int] = 1, padding: int | ~typing.Tuple[int, int] | str | None = None, groups: int = 1, norm_layer: ~typing.Callable[[...], ~torch.nn.modules.module.Module] | None = functools.partial(<class 'torch.nn.modules.batchnorm.BatchNorm2d'>, eps=0.001, momentum=0.99), activation_layer: ~typing.Callable[[...], ~torch.nn.modules.module.Module] | None = <class 'torch.nn.modules.activation.ReLU'>, dilation: int | ~typing.Tuple[int, int] = 1, inplace: bool | None = True, bias: bool | None = None)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.Conv2dSamePadding(in_channels, out_channels, kernel_size, stride, padding=None, dilation=1, groups=1, bias=False)

Part of uvq model.

Immitate the padding=”same” tensorflow convolutions.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.Conv3DSamePadding(in_channels, out_channels, kernel_size, stride=1, dilation=1, groups=1, bias=True)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.InceptionB0B3(conv1_in, conv1_out, pool_shape, include_maxpool=False)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.InceptionB1B2(conv1_in, conv1_out, conv2_out, pool_shape)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.InceptionMixed(conv1_in, b0_conv1_out, b0_pool_shape, b1_conv1_out, b1_conv2_out, b1_pool_shape, b2_conv1_out, b2_conv2_out, b2_pool_shape, b3_conv1_out, b3_pool_shape)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.InceptionMixedBlock

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.Interpolate(size=None, scale_factor=None, mode='bilinear', align_corners=False)

Imitate the BilinearResize layer of tensorflow.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.MBConvSamePadding(input_channels: int, expand_ratio: int, out_channels: int, kernel: int, stride: int, stochastic_depth_prob: float, norm_layer: ~typing.Callable[[...], ~torch.nn.modules.module.Module] = functools.partial(<class 'torch.nn.modules.batchnorm.BatchNorm2d'>, eps=0.001, momentum=0.99), se_layer: ~typing.Callable[[...], ~torch.nn.modules.module.Module] = <class 'torchvision.ops.misc.SqueezeExcitation'>)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.MaxPool3dSame(kernel_size, stride)

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

class cutcutcodec.core.analysis.video.metric.uvq_google.custom_nn_layers.PermuteLayerNHWC(permutation=(0, 2, 3, 1))

Part of uvq model.

Initialize internal Module state, shared by both nn.Module and ScriptModule.