Convolutional Layers

DL4J equivalents and API reference for Keras convolutional layers — Conv1D, Conv2D, Conv3D, SeparableConv2D, transposed convolutions, cropping, upsampling, and zero-padding.

Keras Convolutional Layer Import

Convolutional layer support is implemented in the layers/convolutional package.

Support Summary

Keras Layer	DL4J Equivalent	Supported
Conv1D	Convolution1DLayer	Yes
Conv2D	ConvolutionLayer	Yes
Conv3D	ConvolutionLayer (3D)	Yes
AtrousConvolution1D	Convolution1DLayer (with dilation)	Yes
AtrousConvolution2D	ConvolutionLayer (with dilation)	Yes
SeparableConv1D	—	No
SeparableConv2D	SeparableConvolution2DLayer	Yes
DepthwiseConv2D	DepthwiseConvolution2DLayer	Yes
Conv2DTranspose	Deconvolution2DLayer	Yes
Conv3DTranspose	—	No
Cropping1D	Cropping1DLayer	Yes
Cropping2D	Cropping2DLayer	Yes
Cropping3D	Cropping3DLayer	Yes
UpSampling1D	Upsampling1DLayer	Yes
UpSampling2D	Upsampling2DLayer	Yes
UpSampling3D	Upsampling3DLayer	Yes
ZeroPadding1D	ZeroPadding1DLayer	Yes
ZeroPadding2D	ZeroPaddingLayer	Yes
ZeroPadding3D	ZeroPadding3DLayer	Yes

---

KerasConvolution2D

source

Imports a Keras Conv2D layer as a DL4J ConvolutionLayer.

Constructor

public KerasConvolution2D(Integer kerasVersion) throws UnsupportedKerasConfigurationException

Pass-through constructor. kerasVersion is the major Keras version (1 or 2).

getConvolution2DLayer

public ConvolutionLayer getConvolution2DLayer()

Returns the DL4J ConvolutionLayer.

getOutputType

public InputType getOutputType(InputType... inputType) throws InvalidKerasConfigurationException

---

KerasConvolution1D

source

Imports a Keras Conv1D layer as a DL4J Convolution1DLayer.

Constructor

public KerasConvolution1D(Integer kerasVersion) throws UnsupportedKerasConfigurationException

getConvolution1DLayer

public Convolution1DLayer getConvolution1DLayer()

getInputPreprocessor

public InputPreProcessor getInputPreprocessor(InputType... inputType)
        throws InvalidKerasConfigurationException

setWeights

public void setWeights(Map<String, INDArray> weights) throws InvalidKerasConfigurationException

---

KerasConvolution3D

source

Imports a Keras Conv3D layer as a DL4J 3D ConvolutionLayer.

---

KerasAtrousConvolution1D / KerasAtrousConvolution2D

source 1D | source 2D

Imports Keras 1.x AtrousConvolution (dilated convolution) layers. In Keras 2, dilation is specified via the dilation_rate argument of standard Conv layers; the importer handles both representations.

public KerasAtrousConvolution2D(Integer kerasVersion) throws UnsupportedKerasConfigurationException

---

KerasSeparableConvolution2D

source

Imports a Keras SeparableConv2D layer as a DL4J SeparableConvolution2DLayer. Depthwise and pointwise weight tensors are loaded separately from the HDF5 file.

public KerasSeparableConvolution2D(Integer kerasVersion) throws UnsupportedKerasConfigurationException

---

KerasDeconvolution2D (Conv2DTranspose)

source

Imports a Keras Conv2DTranspose layer as a DL4J Deconvolution2DLayer.

public KerasDeconvolution2D(Integer kerasVersion) throws UnsupportedKerasConfigurationException

---

Cropping Layers

KerasCropping1D

source

public KerasCropping1D(Map<String, Object> layerConfig)
        throws InvalidKerasConfigurationException, UnsupportedKerasConfigurationException

KerasCropping2D

source

public KerasCropping2D(Map<String, Object> layerConfig)
        throws InvalidKerasConfigurationException, UnsupportedKerasConfigurationException

public Cropping2D getCropping2DLayer()

KerasCropping3D

source

---

Upsampling Layers

KerasUpsampling1D

source

public KerasUpsampling1D(Map<String, Object> layerConfig)
        throws InvalidKerasConfigurationException, UnsupportedKerasConfigurationException

public Upsampling1D getUpsampling1DLayer()

KerasUpsampling2D

source

public Upsampling2D getUpsampling2DLayer()

KerasUpsampling3D

source

public Upsampling3D getUpsampling3DLayer()

---

Zero Padding Layers

KerasZeroPadding1D

source

public KerasZeroPadding1D(Map<String, Object> layerConfig)
        throws InvalidKerasConfigurationException, UnsupportedKerasConfigurationException

KerasZeroPadding2D

source

public ZeroPaddingLayer getZeroPadding2DLayer()

KerasZeroPadding3D

source

public ZeroPadding3DLayer getZeroPadding3DLayer()

---

Notes

• Data format: Keras defaults to channels-last (NHWC). DL4J uses channels-first (NCHW) internally. The importer transposes weight tensors during loading so you do not need to manually convert weights.

• SeparableConv1D: not supported. Use Conv1D with groups if available in your Keras version, or restructure the model.

• Conv3DTranspose: no DL4J equivalent at this time.

• Dilation: both the legacy AtrousConvolution names (Keras 1) and the dilation_rate argument (Keras 2) on standard Conv layers are handled transparently.