Package smile.vision

Class EfficientNet

java.lang.Object

smile.deep.layer.LayerBlock

smile.vision.EfficientNet

All Implemented Interfaces:

Function<Tensor,Tensor>, Layer

public class EfficientNet
extends LayerBlock

EfficientNet is an image classification model family. It was first described in EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.

Constructor Summary

Constructors

Constructor	Description
EfficientNet(MBConvConfig[] invertedResidualSetting, double dropout, double stochasticDepthProb, int numClasses, int lastChannel, IntFunction<Layer> normLayer)	Constructor.

Method Summary

Modifier and Type	Method	Description
SequentialBlock	features()	Returns the feature layer block.
Tensor	forward(Tensor input)	Forward propagation (or forward pass) through the layer.
static VisionModel	V2L()	EfficientNet-V2_L (largest) model.
static VisionModel	V2L(String path)	EfficientNet-V2_L (largest) model.
static VisionModel	V2M()	EfficientNet-V2_M (larger) model.
static VisionModel	V2M(String path)	EfficientNet-V2_M (larger) model.
static VisionModel	V2S()	EfficientNet-V2_S (baseline) model.
static VisionModel	V2S(String path)	EfficientNet-V2_S (baseline) model.

Methods inherited from class smile.deep.layer.LayerBlock

add, add, asTorch, toString

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface java.util.function.Function

andThen, compose

Methods inherited from interface smile.deep.layer.Layer

apply, isTraining

Constructor Details

EfficientNet

public EfficientNet(MBConvConfig[] invertedResidualSetting,
 double dropout,
 double stochasticDepthProb,
 int numClasses,
 int lastChannel,
 IntFunction<Layer> normLayer)

Constructor.

Parameters:

invertedResidualSetting - the network structure.

dropout - the dropout probability.

stochasticDepthProb - the stochastic depth probability.

numClasses - the number of classes.

lastChannel - the number of channels on the penultimate layer.

normLayer - the functor to create the normalization layer.

Method Details

forward

public Tensor forward(Tensor input)

Description copied from interface: Layer

Forward propagation (or forward pass) through the layer.

Parameters:

input - the input tensor.

Returns:

the output tensor.

features

public SequentialBlock features()

Returns the feature layer block.

Returns:

the feature layer block.

V2S

public static VisionModel V2S()

EfficientNet-V2_S (baseline) model.

Returns:

the model.

V2S

public static VisionModel V2S(String path)

EfficientNet-V2_S (baseline) model.

Parameters:

path - the pre-trained model file path.

Returns:

the model.

V2M

public static VisionModel V2M()

EfficientNet-V2_M (larger) model.

Returns:

the model.

V2M

public static VisionModel V2M(String path)

EfficientNet-V2_M (larger) model.

Parameters:

path - the pre-trained model file path.

Returns:

the model.

V2L

public static VisionModel V2L()

EfficientNet-V2_L (largest) model.

Returns:

the model.

V2L

public static VisionModel V2L(String path)

EfficientNet-V2_L (largest) model.

Parameters:

path - the pre-trained model file path.

Returns:

the model.