Package ai.djl.nn.convolutional
Contains classes that define convolutional operations extending
Convolution
and Deconvolution
.-
Class Summary Class Description Conv1d AConv1d
layer works similar toConvolution
layer with the exception of the number of dimension it operates on being only one, which isLayoutType.WIDTH
.Conv1d.Builder Conv1dTranspose AConv1dTranspose
layer works similar toDeconvolution
layer with the exception of the number of dimension it operates on being only one, which isLayoutType.WIDTH
.Conv1dTranspose.Builder The Builder to construct aConv1dTranspose
type ofBlock
.Conv2d Being the pioneer of convolution layers,Conv2d
layer works on two dimensions of input,LayoutType.WIDTH
andLayoutType.HEIGHT
as usually aConv2d
layer is used to process data with two spatial dimensions, namely image.Conv2d.Builder Conv2dTranspose Conv2dTranspose.Builder The Builder to construct aConv2dTranspose
type ofBlock
.Conv3d Conv3d
layer behaves just asConvolution
does, with the distinction being it operates of 3-dimensional data such as medical images or video data.Conv3d.Builder Convolution A convolution layer does a dot product calculation on each channel of \(k\)-channel input data by specified number of filters, each containing \(k\) kernels for calculating each channel in the input data and then summed per filter, hence the number of filters denote the number of output channels of a convolution layer.Convolution.ConvolutionBuilder<T extends Convolution.ConvolutionBuilder> A builder that can build anyConvolution
block.Deconvolution Transposed convolution, also named fractionally-strided convolution Dumoulin & Visin or deconvolution Long et al., 2015, serves this purpose.Deconvolution.DeconvolutionBuilder<T extends Deconvolution.DeconvolutionBuilder> A builder that can build anyDeconvolution
block.