Efficient Channel Attention

class complextorch.nn.modules.attention.eca.CVEfficientChannelAttention1d(channels: int, MaskingClass: ~torch.nn.modules.module.Module = <class 'complextorch.nn.modules.mask.ComplexRatioMask'>, b: int = 1, gamma: int = 2)

1-D Complex-Valued Efficient Channel Attention

Implements the operation:

\[\texttt{CV-ECA}(\mathbf{z}) = \mathcal{M}(\text{conv}(H_\texttt{CVAdaptiveAvgPool1d}(\mathbf{z}))) \odot \mathbf{z},\]

where \(\mathcal{M}(\cdot)\) is the masking function (by default, ComplexRatioMask is used) and \(H_\texttt{CVAdaptiveAvgPoolNd}(\cdot)\) is the complex-valued global pooling operator.

Based on work from the following paper:

Q Wang, B Wu, P Zhu, P Li, W Zuo, and Q Hu: ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

• Fig. 2

• https://openaccess.thecvf.com/content_CVPR_2020/papers/Wang_ECA-Net_Efficient_Channel_Attention_for_Deep_Convolutional_Neural_Networks_CVPR_2020_paper.pdf

class complextorch.nn.modules.attention.eca.CVEfficientChannelAttention2d(channels: int, MaskingClass: ~torch.nn.modules.module.Module = <class 'complextorch.nn.modules.mask.ComplexRatioMask'>, b: int = 1, gamma: int = 2)

2-D Complex-Valued Efficient Channel Attention

Implements the operation:

\[\texttt{CV-ECA}(\mathbf{z}) = \mathcal{M}(\text{conv}(H_\texttt{CVAdaptiveAvgPool2d}(\mathbf{z}))) \odot \mathbf{z},\]

where \(\mathcal{M}(\cdot)\) is the masking function (by default, ComplexRatioMask is used) and \(H_\texttt{CVAdaptiveAvgPoolNd}(\cdot)\) is the complex-valued global pooling operator.

Based on work from the following paper:

Q Wang, B Wu, P Zhu, P Li, W Zuo, and Q Hu: ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

• Fig. 2

• https://openaccess.thecvf.com/content_CVPR_2020/papers/Wang_ECA-Net_Efficient_Channel_Attention_for_Deep_Convolutional_Neural_Networks_CVPR_2020_paper.pdf

class complextorch.nn.modules.attention.eca.CVEfficientChannelAttention3d(channels: int, MaskingClass: ~torch.nn.modules.module.Module = <class 'complextorch.nn.modules.mask.ComplexRatioMask'>, b: int = 1, gamma: int = 2)

3-D Complex-Valued Efficient Channel Attention

Implements the operation:

\[\texttt{CV-ECA}(\mathbf{z}) = \mathcal{M}(\text{conv}(H_\texttt{CVAdaptiveAvgPool3d}(\mathbf{z}))) \odot \mathbf{z},\]

where \(\mathcal{M}(\cdot)\) is the masking function (by default, ComplexRatioMask is used) and \(H_\texttt{CVAdaptiveAvgPoolNd}(\cdot)\) is the complex-valued global pooling operator.

Based on work from the following paper:

Q Wang, B Wu, P Zhu, P Li, W Zuo, and Q Hu: ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks

• Fig. 2

• https://openaccess.thecvf.com/content_CVPR_2020/papers/Wang_ECA-Net_Efficient_Channel_Attention_for_Deep_Convolutional_Neural_Networks_CVPR_2020_paper.pdf