weka.dl4j.interpretability

Class AbstractCNNSaliencyMapGenerator

java.lang.Object

weka.dl4j.interpretability.AbstractCNNSaliencyMapGenerator

Direct Known Subclasses:

ScoreCAM

public abstract class AbstractCNNSaliencyMapGenerator
extends java.lang.Object

Generic class for a saliency map generator. Only implemented by ScoreCAM currently so some fields may need to be refactored when implementing a new generator method. This method should be kept as free of Weka classes as possible, so that this can all be refactored into the DL4J package (which doesn't have WEKA as a dependency).

Constructor Summary

Constructors

Constructor and Description
AbstractCNNSaliencyMapGenerator()

Method Summary

Modifier and Type	Method and Description
void	addIterationIncrementListener(IterationIncrementListener listener) Add an event listener to the iterations increment event.
void	addIterationsFinishedListeners(IterationsFinishedListener listener) Add an event listener to the iterations finished event.
void	addIterationsStartedListener(IterationsStartedListener listener) Add an event listener to the iterations started event.
abstract java.awt.image.BufferedImage	generateHeatmapToImage(int[] targetClasses, java.lang.String[] classMap, boolean normalize) Generates heatmaps for the supplied classes, returning a human-viewable heatmap summary.
int	getBatchSize()
org.deeplearning4j.nn.graph.ComputationGraph	getComputationGraph()
org.nd4j.linalg.dataset.api.preprocessor.ImagePreProcessingScaler	getImagePreProcessingScaler()
java.lang.String	getInputFilename()
org.deeplearning4j.nn.conf.inputs.InputType.InputTypeConvolutional	getModelInputShape()
java.lang.String	getModelName()
boolean	isImageChannelsLast()
abstract void	processImage(java.io.File imageFile) Main processing entrypoint.
void	setBatchSize(int batchSize)
void	setComputationGraph(org.deeplearning4j.nn.graph.ComputationGraph computationGraph)
void	setImageChannelsLast(boolean imageChannelsLast)
void	setImagePreProcessingScaler(org.nd4j.linalg.dataset.api.preprocessor.ImagePreProcessingScaler imagePreProcessingScaler)
void	setInputFilename(java.lang.String inputFilename)
void	setModelInputShape(org.deeplearning4j.nn.conf.inputs.InputType.InputTypeConvolutional modelInputShape)
void	setModelName(java.lang.String modelName)

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

AbstractCNNSaliencyMapGenerator

public AbstractCNNSaliencyMapGenerator()

Method Detail

processImage

public abstract void processImage(java.io.File imageFile)

Main processing entrypoint. Takes an image as input, and performs the necessary processing on it. Note that this is performed separately to the actual saliency map generation. This method should only perform the expensive one-time processing on an image.

Parameters:

imageFile - File to process into saliency map.

generateHeatmapToImage

public abstract java.awt.image.BufferedImage generateHeatmapToImage(int[] targetClasses,
                                                                    java.lang.String[] classMap,
                                                                    boolean normalize)

Generates heatmaps for the supplied classes, returning a human-viewable heatmap summary.

Parameters:

targetClasses - Classes to generate saliency maps for.

classMap - Appropriate class map for the supplied classes.

normalize - Should we normalize the heatmap?

Returns:

Human-viewable heatmap summary.

addIterationsStartedListener

public void addIterationsStartedListener(IterationsStartedListener listener)

Add an event listener to the iterations started event.

Parameters:

listener - Listener to call when iterations start.

addIterationIncrementListener

public void addIterationIncrementListener(IterationIncrementListener listener)

Add an event listener to the iterations increment event.

Parameters:

listener - Listener to call when iterations increment.

addIterationsFinishedListeners

public void addIterationsFinishedListeners(IterationsFinishedListener listener)

Add an event listener to the iterations finished event.

Parameters:

listener - Listener to call when iterations finish.

getComputationGraph

public org.deeplearning4j.nn.graph.ComputationGraph getComputationGraph()

setComputationGraph

public void setComputationGraph(org.deeplearning4j.nn.graph.ComputationGraph computationGraph)

getBatchSize

public int getBatchSize()

setBatchSize

public void setBatchSize(int batchSize)

getImagePreProcessingScaler

public org.nd4j.linalg.dataset.api.preprocessor.ImagePreProcessingScaler getImagePreProcessingScaler()

setImagePreProcessingScaler

public void setImagePreProcessingScaler(org.nd4j.linalg.dataset.api.preprocessor.ImagePreProcessingScaler imagePreProcessingScaler)

isImageChannelsLast

public boolean isImageChannelsLast()

setImageChannelsLast

public void setImageChannelsLast(boolean imageChannelsLast)

getModelInputShape

public org.deeplearning4j.nn.conf.inputs.InputType.InputTypeConvolutional getModelInputShape()

setModelInputShape

public void setModelInputShape(org.deeplearning4j.nn.conf.inputs.InputType.InputTypeConvolutional modelInputShape)

getModelName

public java.lang.String getModelName()

setModelName

public void setModelName(java.lang.String modelName)

getInputFilename

public java.lang.String getInputFilename()

setInputFilename

public void setInputFilename(java.lang.String inputFilename)