Package ai.djl.modality

Class Classifications

java.lang.Object

ai.djl.modality.Classifications

All Implemented Interfaces:

BytesSupplier, Ensembleable<Classifications>, ai.djl.util.JsonSerializable, java.io.Serializable

Direct Known Subclasses:

DetectedObjects

public class Classifications
extends java.lang.Object
implements ai.djl.util.JsonSerializable, Ensembleable<Classifications>

Classifications is the container that stores the classification results for classification on a single input.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	Classifications.Classification	A Classification stores the classification result for a single class on a single input.
static class	Classifications.ClassificationsSerializer	A customized Gson serializer to serialize the Classifications object.

Field Summary

Fields

Modifier and Type	Field	Description
protected java.util.List<java.lang.String>	classNames
protected java.util.List<java.lang.Double>	probabilities

Constructor Summary

Constructors

Constructor	Description
Classifications(java.util.List<java.lang.String> classNames, NDArray probabilities)	Constructs a Classifications using list of classNames parallel to an NDArray of probabilities.
Classifications(java.util.List<java.lang.String> classNames, NDArray probabilities, int topK)	Constructs a Classifications using list of classNames parallel to an NDArray of probabilities.
Classifications(java.util.List<java.lang.String> classNames, java.util.List<java.lang.Double> probabilities)	Constructs a Classifications using a parallel list of classNames and probabilities.

Method Summary

Modifier and Type	Method	Description
<T extends Classifications.Classification> T	best()	Returns the most likely class for the classification.
Classifications	ensembleWith(java.util.Iterator<Classifications> it)	Creates an ensembled output with a list of outputs.
<T extends Classifications.Classification> T	get(java.lang.String className)	Returns the result for a particular class name.
java.lang.String	getAsString()	Returns the String presentation of the object.
java.util.List<java.lang.String>	getClassNames()	Returns the classes that were classified into.
java.util.List<java.lang.Double>	getProbabilities()	Returns the list of probabilities for each class (matching the order of the class names).
<T extends Classifications.Classification> T	item(int index)	Returns the item at a given index based on the order used to construct the Classifications.
<T extends Classifications.Classification> java.util.List<T>	items()	Returns a classification item for each potential class for the input.
void	setTopK(int topK)	Set the topK number of classes to be displayed.
java.nio.ByteBuffer	toByteBuffer()	Returns the ByteBuffer presentation of the object.
java.lang.String	toJson()
<T extends Classifications.Classification> java.util.List<T>	topK()	Returns a list of the top classes.
<T extends Classifications.Classification> java.util.List<T>	topK(int k)	Returns a list of the top k best classes.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Methods inherited from interface ai.djl.ndarray.BytesSupplier

getAsBytes, getAsObject

Field Detail

classNames

protected java.util.List<java.lang.String> classNames

probabilities

protected java.util.List<java.lang.Double> probabilities

Constructor Detail

Classifications

public Classifications(java.util.List<java.lang.String> classNames,
                       java.util.List<java.lang.Double> probabilities)

Constructs a Classifications using a parallel list of classNames and probabilities.

Parameters:

classNames - the names of the classes

probabilities - the probabilities for each class for the input

Classifications

public Classifications(java.util.List<java.lang.String> classNames,
                       NDArray probabilities)

Constructs a Classifications using list of classNames parallel to an NDArray of probabilities.

Parameters:

classNames - the names of the classes

probabilities - the probabilities for each class for the input

Classifications

public Classifications(java.util.List<java.lang.String> classNames,
                       NDArray probabilities,
                       int topK)

Constructs a Classifications using list of classNames parallel to an NDArray of probabilities.

Parameters:

classNames - the names of the classes

probabilities - the probabilities for each class for the input

topK - the number of top classes to return

Method Detail

getClassNames

public java.util.List<java.lang.String> getClassNames()

Returns the classes that were classified into.

Returns:

the classes that were classified into

getProbabilities

public java.util.List<java.lang.Double> getProbabilities()

Returns the list of probabilities for each class (matching the order of the class names).

Returns:

the list of probabilities for each class (matching the order of the class names)

setTopK

public final void setTopK(int topK)

Set the topK number of classes to be displayed.

Parameters:

topK - the number of top classes to return

items

public <T extends Classifications.Classification> java.util.List<T> items()

Returns a classification item for each potential class for the input.

Type Parameters:

T - the type of classification item for the task

Returns:

the list of classification items

item

public <T extends Classifications.Classification> T item(int index)

Returns the item at a given index based on the order used to construct the Classifications.

Type Parameters:

T - the type of classification item for the task

Parameters:

index - the index of the item to return

Returns:

the item at the given index, equivalent to classifications.items().get(index)

topK

public <T extends Classifications.Classification> java.util.List<T> topK()

Returns a list of the top classes.

Type Parameters:

T - the type of the classification item for the task

Returns:

the list of classification items for the best classes in order of best to worst

topK

public <T extends Classifications.Classification> java.util.List<T> topK(int k)

Returns a list of the top k best classes.

Type Parameters:

T - the type of the classification item for the task

Parameters:

k - the number of classes to return

Returns:

the list of classification items for the best classes in order of best to worst

best

public <T extends Classifications.Classification> T best()

Returns the most likely class for the classification.

Type Parameters:

T - the type of the classification item for the task

Returns:

the classification item

get

public <T extends Classifications.Classification> T get(java.lang.String className)

Returns the result for a particular class name.

Type Parameters:

T - the type of the classification item for the task

Parameters:

className - the class name to get results for

Returns:

the (first if multiple) classification item

toJson

public java.lang.String toJson()

Specified by:

toJson in interface ai.djl.util.JsonSerializable

getAsString

public java.lang.String getAsString()

Returns the String presentation of the object.

Specified by:

getAsString in interface BytesSupplier

Returns:

the String presentation of the object

toByteBuffer

public java.nio.ByteBuffer toByteBuffer()

Returns the ByteBuffer presentation of the object.

Specified by:

toByteBuffer in interface BytesSupplier

Returns:

the ByteBuffer presentation of the object

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

ensembleWith

public Classifications ensembleWith(java.util.Iterator<Classifications> it)

Creates an ensembled output with a list of outputs.

Specified by:

ensembleWith in interface Ensembleable<Classifications>

Parameters:

it - the outputs to ensemble with. It uses the caller class to determine how to ensemble.

Returns:

the ensembled (averaged) output