Save the date! Google I/O returns May 18-20 Register now

Interpreter

public final class Interpreter

Driver class to drive model inference with TensorFlow Lite.

A Interpreter encapsulates a pre-trained TensorFlow Lite model, in which operations are executed for model inference.

For example, if a model takes only one input and returns only one output: 

try (Interpreter interpreter = new Interpreter(file_of_a_tensorflowlite_model)) {
   interpreter.run(input, output);
 
 }

If a model takes multiple inputs or outputs: 

Object[] inputs = {input0, input1, ...;
 Map map_of_indices_to_outputs = new HashMap<>();
 FloatBuffer ith_output = FloatBuffer.allocateDirect(3 * 2 * 4);  // Float tensor, shape 3x2x4.
 ith_output.order(ByteOrder.nativeOrder());
 map_of_indices_to_outputs.put(i, ith_output);
 try (Interpreter interpreter = new Interpreter(file_of_a_tensorflowlite_model)) {
   interpreter.runForMultipleInputsOutputs(inputs, map_of_indices_to_outputs);
 }
 }

If a model takes or produces string tensors: 

String[] input = {"foo", "bar";  // Input tensor shape is [2].
 String[] output = new String[3][2];  // Output tensor shape is [3, 2].
 try (Interpreter interpreter = new Interpreter(file_of_a_tensorflowlite_model)) {
   interpreter.runForMultipleInputsOutputs(input, output);
 }
 }

Orders of inputs and outputs are determined when converting TensorFlow model to TensorFlowLite model with Toco, as are the default shapes of the inputs.

When inputs are provided as (multi-dimensional) arrays, the corresponding input tensor(s) will be implicitly resized according to that array's shape. When inputs are provided as ERROR(/Buffer) types, no implicit resizing is done; the caller must ensure that the ERROR(/Buffer) byte size either matches that of the corresponding tensor, or that they first resize the tensor via ERROR(/#resizeInput()) . Tensor shape and type information can be obtained via the Tensor class, available via getInputTensor(int) and getOutputTensor(int) .

WARNING: Instances of a Interpreter is not thread-safe. A Interpreter owns resources that must be explicitly freed by invoking close()

The TFLite library is built against NDK API 19. It may work for Android API levels below 19, but is not guaranteed.

Nested Classes

class Interpreter.Options An options class for controlling runtime interpreter behavior.

Public Constructors

Interpreter (File modelFile)
Initializes a Interpreter
Interpreter (File modelFile, int numThreads)
This constructor was deprecated in API level . Prefer using the Interpreter(File, Options) constructor. This method will be removed in a future release.
Interpreter (File modelFile, Interpreter.Options options)
Initializes a Interpreter and specifies the number of threads used for inference.
Interpreter (ByteBuffer byteBuffer)
Initializes a Interpreter with a ByteBuffer of a model file.
Interpreter (ByteBuffer byteBuffer, int numThreads)
This constructor was deprecated in API level . Prefer using the Interpreter(ByteBuffer, Options) constructor. This method will be removed in a future release.
Interpreter (MappedByteBuffer mappedByteBuffer)
This constructor was deprecated in API level . Prefer using the Interpreter(ByteBuffer, Options) constructor. This method will be removed in a future release.
Interpreter (ByteBuffer byteBuffer, Interpreter.Options options)
Initializes a Interpreter with a ByteBuffer of a model file and a set of custom ERROR(/#Options) .

Public Methods

void
allocateTensors ()
Expicitly updates allocations for all tensors, if necessary.
void
close ()
Release resources associated with the Interpreter .
int
getInputIndex (String opName)
Gets index of an input given the op name of the input.
Tensor
getInputTensor (int inputIndex)
Gets the Tensor associated with the provdied input index.
int
getInputTensorCount ()
Gets the number of input tensors.
Long
getLastNativeInferenceDurationNanoseconds ()
Returns native inference timing.
int
getOutputIndex (String opName)
Gets index of an output given the op name of the output.
Tensor
getOutputTensor (int outputIndex)
Gets the Tensor associated with the provdied output index.
int
getOutputTensorCount ()
Gets the number of output Tensors.
void
modifyGraphWithDelegate ( Delegate delegate)
Advanced: Modifies the graph with the provided Delegate .
void
resetVariableTensors ()
Advanced: Resets all variable tensors to the default value.
void
resizeInput (int idx, int[] dims, boolean strict)
Resizes idx-th input of the native model to the given dims.
void
resizeInput (int idx, int[] dims)
Resizes idx-th input of the native model to the given dims.
void
run (Object input, Object output)
Runs model inference if the model takes only one input, and provides only one output.
void
runForMultipleInputsOutputs (Object[] inputs, Map<Integer, Object> outputs)
Runs model inference if the model takes multiple inputs, or returns multiple outputs.
void
setCancelled (boolean cancelled)
Advanced: Interrupts inference in the middle of a call to run(Object, Object) .
void
setNumThreads (int numThreads)
This method was deprecated in API level . Prefer using setNumThreads(int) directly for controlling thread multi-threading. This method will be removed in a future release.

Inherited Methods

Public Constructors

public Interpreter (File modelFile)

Initializes a Interpreter

Throws
IllegalArgumentException if modelFile does not encode a valid TensorFlow Lite model.

public Interpreter (File modelFile, int numThreads)

This constructor was deprecated in API level .
Prefer using the Interpreter(File, Options) constructor. This method will be removed in a future release.

Initializes a Interpreter and specifies the number of threads used for inference.

public Interpreter (File modelFile, Interpreter.Options options)

Initializes a Interpreter and specifies the number of threads used for inference.

Throws
IllegalArgumentException if modelFile does not encode a valid TensorFlow Lite model.

public Interpreter (ByteBuffer byteBuffer)

Initializes a Interpreter with a ByteBuffer of a model file.

The ByteBuffer should not be modified after the construction of a Interpreter . The ByteBuffer can be either a MappedByteBuffer that memory-maps a model file, or a direct ByteBuffer of nativeOrder() that contains the bytes content of a model.

Throws
IllegalArgumentException if byteBuffer is not a MappedByteBuffer nor a direct ERROR(/Bytebuffer) of nativeOrder.

public Interpreter (ByteBuffer byteBuffer, int numThreads)

This constructor was deprecated in API level .
Prefer using the Interpreter(ByteBuffer, Options) constructor. This method will be removed in a future release.

Initializes a Interpreter with a ByteBuffer of a model file and specifies the number of threads used for inference.

The ByteBuffer should not be modified after the construction of a Interpreter . The ByteBuffer can be either a MappedByteBuffer that memory-maps a model file, or a direct ByteBuffer of nativeOrder() that contains the bytes content of a model.

public Interpreter (MappedByteBuffer mappedByteBuffer)

This constructor was deprecated in API level .
Prefer using the Interpreter(ByteBuffer, Options) constructor. This method will be removed in a future release.

Initializes a Interpreter with a MappedByteBuffer to the model file.

The MappedByteBuffer should remain unchanged after the construction of a Interpreter .

public Interpreter (ByteBuffer byteBuffer, Interpreter.Options options)

Initializes a Interpreter with a ByteBuffer of a model file and a set of custom ERROR(/#Options) .

The ByteBuffer should not be modified after the construction of a Interpreter . The ByteBuffer can be either a MappedByteBuffer that memory-maps a model file, or a direct ByteBuffer of nativeOrder() that contains the bytes content of a model.

Throws
IllegalArgumentException if byteBuffer is not a MappedByteBuffer nor a direct ERROR(/Bytebuffer) of nativeOrder.

Public Methods

public void allocateTensors ()

Expicitly updates allocations for all tensors, if necessary.

This will propagate shapes and memory allocations for all dependent tensors using the input tensor shape(s) as given.

Note: This call is *purely optional*. Tensor allocation will occur automatically during execution if any input tensors have been resized. This call is most useful in determining the shapes for any output tensors before executing the graph, e.g., 

interpreter.resizeInput(0, new int[]{1, 4, 4, 3));
 interpreter.allocateTensors();
 FloatBuffer input = FloatBuffer.allocate(interpreter.getInputTensor(0),numElements());
 // Populate inputs...
 FloatBuffer output = FloatBuffer.allocate(interpreter.getOutputTensor(0).numElements());
 interpreter.run(input, output)
 // Process outputs...
 }

Throws
IllegalStateException if the graph's tensors could not be successfully allocated.

public void close ()

Release resources associated with the Interpreter .

public int getInputIndex (String opName)

Gets index of an input given the op name of the input.

Throws
IllegalArgumentException if opName does not match any input in the model used to initialize the Interpreter .

public Tensor getInputTensor (int inputIndex)

Gets the Tensor associated with the provdied input index.

Throws
IllegalArgumentException if inputIndex is negtive or is not smaller than the number of model inputs.

public int getInputTensorCount ()

Gets the number of input tensors.

public Long getLastNativeInferenceDurationNanoseconds ()

Returns native inference timing.

Throws
IllegalArgumentException if the model is not initialized by the Interpreter .

public int getOutputIndex (String opName)

Gets index of an output given the op name of the output.

Throws
IllegalArgumentException if opName does not match any output in the model used to initialize the Interpreter .

public Tensor getOutputTensor (int outputIndex)

Gets the Tensor associated with the provdied output index.

Note: Output tensor details (e.g., shape) may not be fully populated until after inference is executed. If you need updated details *before* running inference (e.g., after resizing an input tensor, which may invalidate output tensor shapes), use allocateTensors() to explicitly trigger allocation and shape propagation. Note that, for graphs with output shapes that are dependent on input *values*, the output shape may not be fully determined until running inference.

Throws
IllegalArgumentException if outputIndex is negtive or is not smaller than the number of model outputs.

public int getOutputTensorCount ()

Gets the number of output Tensors.

public void modifyGraphWithDelegate ( Delegate delegate)

Advanced: Modifies the graph with the provided Delegate .

Note: The typical path for providing delegates is via addDelegate(Delegate) , at creation time. This path should only be used when a delegate might require coordinated interaction between Interpeter creation and delegate application.

WARNING: This is an experimental API and subject to change.

Throws
IllegalArgumentException if error occurs when modifying graph with delegate .

public void resetVariableTensors ()

Advanced: Resets all variable tensors to the default value.

If a variable tensor doesn't have an associated buffer, it will be reset to zero.

WARNING: This is an experimental API and subject to change.

public void resizeInput (int idx, int[] dims, boolean strict)

Resizes idx-th input of the native model to the given dims.

When `strict` is True, only unknown dimensions can be resized. Unknown dimensions are indicated as `-1` in the array returned by `Tensor.shapeSignature()`.

Throws
IllegalArgumentException if idx is negtive or is not smaller than the number of model inputs; or if error occurs when resizing the idx-th input. Additionally, the error occurs when attempting to resize a tensor with fixed dimensions when `struct` is True.

public void resizeInput (int idx, int[] dims)

Resizes idx-th input of the native model to the given dims.

Throws
IllegalArgumentException if idx is negtive or is not smaller than the number of model inputs; or if error occurs when resizing the idx-th input.

public void run (Object input, Object output)

Runs model inference if the model takes only one input, and provides only one output.

Warning: The API is more efficient if a ERROR(/Buffer) (preferably direct, but not required) is used as the input/output data type. Please consider using ERROR(/Buffer) to feed and fetch primitive data for better performance. The following concrete ERROR(/Buffer) types are supported:

Note that boolean types are only supported as arrays, not ERROR(/Buffer) s, or as scalar inputs.

Parameters
input an array or multidimensional array, or a ERROR(/Buffer) of primitive types including int, float, long, and byte. ERROR(/Buffer) is the preferred way to pass large input data for primitive types, whereas string types require using the (multi-dimensional) array input path. When a ERROR(/Buffer) is used, its content should remain unchanged until model inference is done, and the caller must ensure that the ERROR(/Buffer) is at the appropriate read position. A null value is allowed only if the caller is using a Delegate that allows buffer handle interop, and such a buffer has been bound to the input Tensor .
output a multidimensional array of output data, or a ERROR(/Buffer) of primitive types including int, float, long, and byte. When a ERROR(/Buffer) is used, the caller must ensure that it is set the appropriate write position. A null value is allowed only if the caller is using a Delegate that allows buffer handle interop, and such a buffer has been bound to the output Tensor . See ERROR(/Options#setAllowBufferHandleOutput()) .
Throws
IllegalArgumentException if input or output is null or empty, or if error occurs when running the inference.
IllegalArgumentException (EXPERIMENTAL, subject to change) if the inference is interrupted by setCancelled(true) .

public void runForMultipleInputsOutputs (Object[] inputs, Map<Integer, Object> outputs)

Runs model inference if the model takes multiple inputs, or returns multiple outputs.

Warning: The API is more efficient if ERROR(/Buffer) s (preferably direct, but not required) are used as the input/output data types. Please consider using ERROR(/Buffer) to feed and fetch primitive data for better performance. The following concrete ERROR(/Buffer) types are supported:

Note that boolean types are only supported as arrays, not ERROR(/Buffer) s, or as scalar inputs.

Note: null values for invididual elements of inputs and outputs is allowed only if the caller is using a Delegate that allows buffer handle interop, and such a buffer has been bound to the corresponding input or output Tensor (s).

Parameters
inputs an array of input data. The inputs should be in the same order as inputs of the model. Each input can be an array or multidimensional array, or a ERROR(/Buffer) of primitive types including int, float, long, and byte. ERROR(/Buffer) is the preferred way to pass large input data, whereas string types require using the (multi-dimensional) array input path. When ERROR(/Buffer) is used, its content should remain unchanged until model inference is done, and the caller must ensure that the ERROR(/Buffer) is at the appropriate read position.
outputs a map mapping output indices to multidimensional arrays of output data or ERROR(/Buffer) s of primitive types including int, float, long, and byte. It only needs to keep entries for the outputs to be used. When a ERROR(/Buffer) is used, the caller must ensure that it is set the appropriate write position.
Throws
IllegalArgumentException if inputs or outputs is null or empty, or if error occurs when running the inference.

public void setCancelled (boolean cancelled)

Advanced: Interrupts inference in the middle of a call to run(Object, Object) .

A cancellation flag will be set to true when this function gets called. The interpreter will check the flag between Op invocations, and if it's true , the interpreter will stop execution. The interpreter will remain a cancelled state until explicitly "uncancelled" by setCancelled(false) .

WARNING: This is an experimental API and subject to change.

Parameters
cancelled true to cancel inference in a best-effort way; false to resume.
Throws
IllegalStateException if the interpreter is not initialized with the cancellable option, which is by default off.
See Also

public void setNumThreads (int numThreads)

This method was deprecated in API level .
Prefer using setNumThreads(int) directly for controlling thread multi-threading. This method will be removed in a future release.

Sets the number of threads to be used for ops that support multi-threading.