io.spokestack.spokestack.tensorflow

Class TensorflowModel

java.lang.Object

io.spokestack.spokestack.tensorflow.TensorflowModel

All Implemented Interfaces:

AutoCloseable

public class TensorflowModel
extends Object
implements AutoCloseable

Tensorflow-Lite model wrapper and loader

This class wraps the TensorFlow Lite interpreter so that it can be mocked for unit testing on non-android platforms. It also encapsulates the input/output byte buffers used for passing input tensors into a model and retrieving outputs.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	TensorflowModel.Loader loader (builder) class for the tensorflow model.

Constructor Summary

Constructors

Constructor and Description
TensorflowModel(TensorflowModel.Loader loader) constructs a new tensorflow model.

Method Summary

Modifier and Type	Method and Description
void	close() releases the tensorflow interpreter.
int	getInputSize()
ByteBuffer	inputs(int index) Get the input buffer at the specified index.
ByteBuffer	outputs(int index) Get the output buffer at the specified index.
void	run() executes the model using the attached buffers.
ByteBuffer	states()

Methods inherited from class java.lang.Object

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

Constructor Detail

TensorflowModel

public TensorflowModel(TensorflowModel.Loader loader)

constructs a new tensorflow model.

Parameters:

loader - the loader (builder) for the model

Method Detail

getInputSize

public int getInputSize()

Returns:

the byte size of the model's inputs.

close

public void close()

releases the tensorflow interpreter.

Specified by:

close in interface AutoCloseable

inputs

public ByteBuffer inputs(int index)

Get the input buffer at the specified index.

Parameters:

index - The index of the desired input buffer.

Returns:

the input tensor buffer at the specified index.

states

public ByteBuffer states()

Returns:

the state tensor buffer

outputs

public ByteBuffer outputs(int index)

Get the output buffer at the specified index.

Parameters:

index - The index of the desired output buffer.

Returns:

the output tensor buffer at the specified index.

run

public void run()

executes the model using the attached buffers.