weka.dl4j.inference

Class ModelOutputDecoder

java.lang.Object

weka.dl4j.inference.ModelOutputDecoder

All Implemented Interfaces:

java.io.Serializable, weka.core.OptionHandler

public class ModelOutputDecoder
extends java.lang.Object
implements java.io.Serializable, weka.core.OptionHandler

Decodes model outputs into a human-readable and more workable format. Holds the class map to be used for decoding

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
ModelOutputDecoder()

Method Summary

Modifier and Type	Method and Description
TopNPredictions	decodePredictions(org.nd4j.linalg.api.ndarray.INDArray predictions) Decode predictions, without any image or model name.
TopNPredictions	decodePredictions(org.nd4j.linalg.api.ndarray.INDArray predictions, java.lang.String imageName, java.lang.String modelName) Main entrypoint - decode the model predictions, saving the image and model name alongside it.
ClassmapType	getBuiltInClassMap()
java.lang.String[]	getClasses() Parses the classmap file into a String[]
java.io.File	getClassMapFile()
java.lang.String[]	getOptions() Gets the current settings of the Classifier.
java.util.Enumeration<weka.core.Option>	listOptions() Returns an enumeration describing the available options.
static java.lang.String[]	parseClassmap(weka.core.Instances instances) Get a classmap from a set of instances.
static java.lang.String[]	parseClassmapFromArff(java.lang.String filepath) Parse a classmap from a valid .arff file.
static java.lang.String[]	parseClassmapFromCsv(java.lang.String filepath) Parse a classmap from a valid .csv dataset.
void	setBuiltInClassMap(ClassmapType classMap)
void	setClassMapFile(java.io.File f)
void	setOptions(java.lang.String[] options) Parses a given list of options.

Methods inherited from class java.lang.Object

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

Methods inherited from interface weka.core.OptionHandler

makeCopy

Constructor Detail

ModelOutputDecoder

public ModelOutputDecoder()

Method Detail

decodePredictions

public TopNPredictions decodePredictions(org.nd4j.linalg.api.ndarray.INDArray predictions)
                                  throws java.lang.Exception

Decode predictions, without any image or model name.

Parameters:

predictions - Predictions to decode

Returns:

Parsed predictions.

Throws:

java.lang.Exception - Exception.

decodePredictions

public TopNPredictions decodePredictions(org.nd4j.linalg.api.ndarray.INDArray predictions,
                                         java.lang.String imageName,
                                         java.lang.String modelName)
                                  throws java.lang.Exception

Main entrypoint - decode the model predictions, saving the image and model name alongside it.

Parameters:

predictions - Predictions to decode

imageName - Name of the image used for prediction

modelName - Name of the model used for prediction

Returns:

TopNPredictions object, parsed from the predictions

Throws:

java.lang.Exception - exception

parseClassmapFromArff

public static java.lang.String[] parseClassmapFromArff(java.lang.String filepath)
                                                throws java.io.IOException

Parse a classmap from a valid .arff file.

Parameters:

filepath - Filepath to arff file

Returns:

Parsed classmap.

Throws:

java.io.IOException - file exception.

parseClassmapFromCsv

public static java.lang.String[] parseClassmapFromCsv(java.lang.String filepath)
                                               throws java.io.IOException

Parse a classmap from a valid .csv dataset.

Parameters:

filepath - Filepath to a valid .csv dataset

Returns:

Parsed classmap.

Throws:

java.io.IOException - file exception.

parseClassmap

public static java.lang.String[] parseClassmap(weka.core.Instances instances)

Get a classmap from a set of instances.

Parameters:

instances - Instances to parse classes from

Returns:

Parsed classmap.

getClasses

public java.lang.String[] getClasses()

Parses the classmap file into a String[]

Returns:

String[], one item for each class

Throws:

java.lang.Exception

getBuiltInClassMap

@OptionMetadata(displayName="Built in class map",
                description="A predefined class map based on a specific dataset (IMAGENET, VGGFACE). Useful when using a pretrained zoo model as these are often trained on IMAGENET.",
                commandLineParamName="builtIn",
                commandLineParamSynopsis="-builtIn {IMAGENET, VGGFACE, CUSTOM}")
public ClassmapType getBuiltInClassMap()

setBuiltInClassMap

public void setBuiltInClassMap(ClassmapType classMap)

getClassMapFile

@OptionMetadata(displayName="Class map file",
                description="File containing mappings from class IDs to human-readable names - can be .txt or .arff",
                commandLineParamName="classMapFile",
                commandLineParamSynopsis="-classMapFile <file location>")
public java.io.File getClassMapFile()

setClassMapFile

public void setClassMapFile(java.io.File f)

listOptions

public java.util.Enumeration<weka.core.Option> listOptions()

Returns an enumeration describing the available options.

Specified by:

listOptions in interface weka.core.OptionHandler

Returns:

an enumeration of all the available options.

getOptions

public java.lang.String[] getOptions()

Gets the current settings of the Classifier.

Specified by:

getOptions in interface weka.core.OptionHandler

Returns:

an array of strings suitable for passing to setOptions

setOptions

public void setOptions(java.lang.String[] options)
                throws java.lang.Exception

Parses a given list of options.

Specified by:

setOptions in interface weka.core.OptionHandler

Parameters:

options - the list of options as an array of strings

Throws:

java.lang.Exception - if an option is not supported