ContextQMD
Back to Tensorflow

lite/c/c_api.h

tensorflow/lite/g3doc/api_docs/c/group/c-api.html

2.21.043.3 KB
Original Source

lite/c/c_api.h

C API for TensorFlow Lite.

Summary

The API leans towards simplicity and uniformity instead of convenience, as most usage will be by language-specific wrappers. It provides largely the same set of functionality as that of the C++ TensorFlow Lite Interpreter API, but is useful for shared libraries where having a stable ABI boundary is important.

Conventions:

  • We use the prefix TfLite for everything in the API.
  • size_t is used to represent byte sizes of objects that are materialized in the address space of the calling process.
  • int is used as an index into arrays.

Usage:

// Create the model and interpreter options.
TfLiteModel* model = TfLiteModelCreateFromFile("/path/to/model.tflite");
TfLiteInterpreterOptions* options = TfLiteInterpreterOptionsCreate();
TfLiteInterpreterOptionsSetNumThreads(options, 2);

// Create the interpreter.
TfLiteInterpreter* interpreter = TfLiteInterpreterCreate(model, options);

// Allocate tensors and populate the input tensor data.
TfLiteInterpreterAllocateTensors(interpreter);
TfLiteTensor* input_tensor =
    TfLiteInterpreterGetInputTensor(interpreter, 0);
TfLiteTensorCopyFromBuffer(input_tensor, input.data(),
                           input.size() * sizeof(float));

// Execute inference.
TfLiteInterpreterInvoke(interpreter);

// Extract the output tensor data.
const TfLiteTensor* output_tensor =
     TfLiteInterpreterGetOutputTensor(interpreter, 0);
TfLiteTensorCopyToBuffer(output_tensor, output.data(),
                         output.size() * sizeof(float));

// Dispose of the model and interpreter objects.
TfLiteInterpreterDelete(interpreter);
TfLiteInterpreterOptionsDelete(options);
TfLiteModelDelete(model);

|

Typedefs

| | --- | | TfLiteInterpreter | typedef struct TfLiteInterpreter

TfLiteInterpreter provides inference from a provided model. | | TfLiteInterpreterOptions | typedef struct TfLiteInterpreterOptions

TfLiteInterpreterOptions allows customized interpreter configuration. | | TfLiteModel | typedef struct TfLiteModel

TfLiteModel wraps a loaded TensorFlow Lite model. | | TfLiteSignatureRunner | typedef struct TfLiteSignatureRunner

TfLiteSignatureRunner is used to run inference on a signature. | | TfLiteTensor | typedef struct TfLiteTensor

A tensor in the interpreter system which is a wrapper around a buffer of data including a dimensionality (or NULL if not currently defined). |

|

Functions

| | --- | | TfLiteExtensionApisVersion(void) | TFL_CAPI_EXPORT const char *

The TensorFlow Lite Extension APIs version. | | TfLiteInterpreterAllocateTensors(TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT TfLiteStatus

Updates allocations for all tensors, resizing dependent tensors using the specified input tensor dimensionality. | | TfLiteInterpreterCancel(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT TfLiteStatus

Tries to cancel any in-flight invocation. | | TfLiteInterpreterCreate(const TfLiteModel *model, const TfLiteInterpreterOptions *optional_options) | TFL_CAPI_EXPORT TfLiteInterpreter *

Returns a new interpreter using the provided model and options, or null on failure. | | TfLiteInterpreterDelete(TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT void

Destroys the interpreter. | | TfLiteInterpreterGetInputTensor(const TfLiteInterpreter *interpreter, int32_t input_index) | TFL_CAPI_EXPORT TfLiteTensor *

Returns the tensor associated with the input index. | | TfLiteInterpreterGetInputTensorCount(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT int32_t

Returns the number of input tensors associated with the model. | | TfLiteInterpreterGetOutputTensor(const TfLiteInterpreter *interpreter, int32_t output_index) | TFL_CAPI_EXPORT const TfLiteTensor *

Returns the tensor associated with the output index. | | TfLiteInterpreterGetOutputTensorCount(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT int32_t

Returns the number of output tensors associated with the model. | | TfLiteInterpreterGetSignatureCount(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT int32_t

SignatureRunner APIs | | TfLiteInterpreterGetSignatureKey(const TfLiteInterpreter *interpreter, int32_t signature_index) | TFL_CAPI_EXPORT const char *

Returns the key of the Nth signature in the model, where N is specified as signature_index. | | TfLiteInterpreterGetSignatureRunner(const TfLiteInterpreter *interpreter, const char *signature_key) | TFL_CAPI_EXPORT TfLiteSignatureRunner *

Returns a new signature runner using the provided interpreter and signature key, or nullptr on failure. | | TfLiteInterpreterGetTensor(const TfLiteInterpreter *interpreter, int index) | TFL_CAPI_EXPORT TfLiteTensor *

Returns modifiable access to the tensor that corresponds to the specified index and is associated with the provided interpreter. | | TfLiteInterpreterInputTensorIndices(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT const int *

Returns a pointer to an array of input tensor indices. | | TfLiteInterpreterInvoke(TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT TfLiteStatus

Runs inference for the loaded graph. | | TfLiteInterpreterOptionsAddDelegate(TfLiteInterpreterOptions *options, TfLiteOpaqueDelegate *delegate) | TFL_CAPI_EXPORT void

Adds a delegate to be applied during TfLiteInterpreter creation. | | TfLiteInterpreterOptionsAddRegistrationExternal(TfLiteInterpreterOptions *options, TfLiteRegistrationExternal *registration) | TFL_CAPI_EXPORT void

Adds an op registration to be applied during TfLiteInterpreter creation. | | TfLiteInterpreterOptionsCopy(const TfLiteInterpreterOptions *from) | TFL_CAPI_EXPORT TfLiteInterpreterOptions *

Creates and returns a shallow copy of an options object. | | TfLiteInterpreterOptionsCreate() | TFL_CAPI_EXPORT TfLiteInterpreterOptions *

Returns a new interpreter options instances. | | TfLiteInterpreterOptionsDelete(TfLiteInterpreterOptions *options) | TFL_CAPI_EXPORT void

Destroys the interpreter options instance. | | TfLiteInterpreterOptionsEnableCancellation(TfLiteInterpreterOptions *options, bool enable) | TFL_CAPI_EXPORT TfLiteStatus

Enables users to cancel in-flight invocations with TfLiteInterpreterCancel. | | TfLiteInterpreterOptionsSetErrorReporter(TfLiteInterpreterOptions *options, void(*)(void *user_data, const char *format, va_list args) reporter, void *user_data) | TFL_CAPI_EXPORT void

Sets a custom error reporter for interpreter execution. | | TfLiteInterpreterOptionsSetNumThreads(TfLiteInterpreterOptions *options, int32_t num_threads) | TFL_CAPI_EXPORT void

Sets the number of CPU threads to use for the interpreter. | | TfLiteInterpreterOutputTensorIndices(const TfLiteInterpreter *interpreter) | TFL_CAPI_EXPORT const int *

Returns a pointer to an array of output tensor indices. | | TfLiteInterpreterResizeInputTensor(TfLiteInterpreter *interpreter, int32_t input_index, const int *input_dims, int32_t input_dims_size) | TFL_CAPI_EXPORT TfLiteStatus

Resizes the specified input tensor. | | TfLiteModelCreate(const void *model_data, size_t model_size) | TFL_CAPI_EXPORT TfLiteModel *

Returns a model from the provided buffer, or null on failure. | | TfLiteModelCreateFromFile(const char *model_path) | TFL_CAPI_EXPORT TfLiteModel *

Returns a model from the provided file, or null on failure. | | TfLiteModelCreateFromFileWithErrorReporter(const char *model_path, void(*)(void *user_data, const char *format, va_list args) reporter, void *user_data) | TFL_CAPI_EXPORT TfLiteModel *

Same as TfLiteModelCreateFromFile with customizble error reporter. | | TfLiteModelCreateWithErrorReporter(const void *model_data, size_t model_size, void(*)(void *user_data, const char *format, va_list args) reporter, void *user_data) | TFL_CAPI_EXPORT TfLiteModel *

Same as TfLiteModelCreate with customizble error reporter. | | TfLiteModelDelete(TfLiteModel *model) | TFL_CAPI_EXPORT void

Destroys the model instance. | | TfLiteSchemaVersion(void) | TFL_CAPI_EXPORT int

The supported TensorFlow Lite model file Schema version. | | TfLiteSignatureRunnerAllocateTensors(TfLiteSignatureRunner *signature_runner) | TFL_CAPI_EXPORT TfLiteStatus

Updates allocations for tensors associated with a signature and resizes dependent tensors using the specified input tensor dimensionality. | | TfLiteSignatureRunnerDelete(TfLiteSignatureRunner *signature_runner) | TFL_CAPI_EXPORT void

Destroys the signature runner. | | TfLiteSignatureRunnerGetInputCount(const TfLiteSignatureRunner *signature_runner) | TFL_CAPI_EXPORT size_t

Returns the number of inputs associated with a signature. | | TfLiteSignatureRunnerGetInputName(const TfLiteSignatureRunner *signature_runner, int32_t input_index) | TFL_CAPI_EXPORT const char *

Returns the (null-terminated) name of the Nth input in a signature, where N is specified as input_index. | | TfLiteSignatureRunnerGetInputTensor(TfLiteSignatureRunner *signature_runner, const char *input_name) | TFL_CAPI_EXPORT TfLiteTensor *

Returns the input tensor identified by input_name in the given signature. | | TfLiteSignatureRunnerGetOutputCount(const TfLiteSignatureRunner *signature_runner) | TFL_CAPI_EXPORT size_t

Returns the number of output tensors associated with the signature. | | TfLiteSignatureRunnerGetOutputName(const TfLiteSignatureRunner *signature_runner, int32_t output_index) | TFL_CAPI_EXPORT const char *

Returns the (null-terminated) name of the Nth output in a signature, where N is specified as output_index. | | TfLiteSignatureRunnerGetOutputTensor(const TfLiteSignatureRunner *signature_runner, const char *output_name) | TFL_CAPI_EXPORT const TfLiteTensor *

Returns the output tensor identified by output_name in the given signature. | | TfLiteSignatureRunnerInvoke(TfLiteSignatureRunner *signature_runner) | TFL_CAPI_EXPORT TfLiteStatus

Runs inference on a given signature. | | TfLiteSignatureRunnerResizeInputTensor(TfLiteSignatureRunner *signature_runner, const char *input_name, const int *input_dims, int32_t input_dims_size) | TFL_CAPI_EXPORT TfLiteStatus

Resizes the input tensor identified as input_name to be the dimensions specified by input_dims and input_dims_size. | | TfLiteTensorByteSize(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT size_t

Returns the size of the underlying data in bytes. | | TfLiteTensorCopyFromBuffer(TfLiteTensor *tensor, const void *input_data, size_t input_data_size) | TFL_CAPI_EXPORT TfLiteStatus

Copies from the provided input buffer into the tensor's buffer. | | TfLiteTensorCopyToBuffer(const TfLiteTensor *output_tensor, void *output_data, size_t output_data_size) | TFL_CAPI_EXPORT TfLiteStatus

Copies to the provided output buffer from the tensor's buffer. | | TfLiteTensorData(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT void *

Returns a pointer to the underlying data buffer. | | TfLiteTensorDim(const TfLiteTensor *tensor, int32_t dim_index) | TFL_CAPI_EXPORT int32_t

Returns the length of the tensor in the "dim_index" dimension. | | TfLiteTensorName(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT const char *

Returns the (null-terminated) name of the tensor. | | TfLiteTensorNumDims(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT int32_t

Returns the number of dimensions that the tensor has. | | TfLiteTensorQuantizationParams(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT TfLiteQuantizationParams

Returns the parameters for asymmetric quantization. | | TfLiteTensorType(const TfLiteTensor *tensor) | TFL_CAPI_EXPORT TfLiteType

Returns the type of a tensor element. | | TfLiteVersion(void) | TFL_CAPI_EXPORT const char *

The TensorFlow Lite Runtime version. |

Typedefs

TfLiteInterpreter

struct[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)TfLiteInterpreter

TfLiteInterpreter provides inference from a provided model.

TfLiteInterpreterOptions

struct[TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)TfLiteInterpreterOptions

TfLiteInterpreterOptions allows customized interpreter configuration.

TfLiteModel

struct[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)TfLiteModel

TfLiteModel wraps a loaded TensorFlow Lite model.

TfLiteSignatureRunner

struct[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)TfLiteSignatureRunner

TfLiteSignatureRunner is used to run inference on a signature.

Note: A signature is used to define a computation in a TF model. A model can have multiple signatures. Each signature contains three components:

  • Signature Key: A unique string to identify a signature
  • Inputs: A list of names, each mapped to an input tensor of a signature
  • Outputs: A list of names, each mapped to an output tensor of a signature

To learn more about signatures in TFLite, refer to: https://www.tensorflow.org/lite/guide/signatures

Using the TfLiteSignatureRunner, for a particular signature, you can set its inputs, invoke (i.e. execute) the computation, and retrieve its outputs.

TfLiteTensor

struct[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)TfLiteTensor

A tensor in the interpreter system which is a wrapper around a buffer of data including a dimensionality (or NULL if not currently defined).

Functions

TfLiteExtensionApisVersion

TFL_CAPI_EXPORT const char * TfLiteExtensionApisVersion(
  void
)

The TensorFlow Lite Extension APIs version.

Returns a pointer to a statically allocated string that is the version number of the TF Lite Extension APIs supported by the (potentially dynamically loaded) TF Lite Runtime library. The TF Lite "Extension APIs" are the APIs for extending TF Lite with custom ops and delegates. More specifically, this version number covers the (non-experimental) functionality documented in the following header files:

  • lite/c/c_api_opaque.h
  • lite/c/common.h
  • lite/c/builtin_op_data.h
  • lite/builtin_ops.h

This version number uses semantic versioning, and the return value should be in semver 2 format http://semver.org, starting with MAJOR.MINOR.PATCH, e.g. "2.14.0" or "2.15.0-rc2".

TfLiteInterpreterAllocateTensors

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteInterpreterAllocateTensors([TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group__c__api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Updates allocations for all tensors, resizing dependent tensors using the specified input tensor dimensionality.

This is a relatively expensive operation, and need only be called after creating the graph and/or resizing any inputs.

TfLiteInterpreterCancel

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteInterpreterCancel(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Tries to cancel any in-flight invocation.

Returns kTfLiteError if cancellation is not enabled via TfLiteInterpreterOptionsEnableCancellation.

TfLiteInterpreterCreate

TFL_CAPI_EXPORT[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)* TfLiteInterpreterCreate(
  const[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)*model,
  const[TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*optional_options
)

Returns a new interpreter using the provided model and options, or null on failure.

  • model must be a valid model instance. The caller retains ownership of the object, and may destroy it (via TfLiteModelDelete) immediately after creating the interpreter. However, if the TfLiteModel was allocated with TfLiteModelCreate, then the model_data buffer that was passed to TfLiteModelCreate must outlive the lifetime of the TfLiteInterpreter object that this function returns, and must not be modified during that time; and if the TfLiteModel was allocated with TfLiteModelCreateFromFile, then the contents of the model file must not be modified during the lifetime of the TfLiteInterpreter object that this function returns.
  • optional_options may be null. The caller retains ownership of the object, and can safely destroy it (via TfLiteInterpreterOptionsDelete) immediately after creating the interpreter.

TfLiteInterpreterDelete

TFL_CAPI_EXPORT void TfLiteInterpreterDelete([TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Destroys the interpreter.

TfLiteInterpreterGetInputTensor

TFL_CAPI_EXPORT[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)* TfLiteInterpreterGetInputTensor(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  int32_t input_index
)

Returns the tensor associated with the input index.

REQUIRES: 0 <= input_index < TfLiteInterpreterGetInputTensorCount(tensor)

TfLiteInterpreterGetInputTensorCount

TFL_CAPI_EXPORT int32_t TfLiteInterpreterGetInputTensorCount(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Returns the number of input tensors associated with the model.

TfLiteInterpreterGetOutputTensor

TFL_CAPI_EXPORT const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)* TfLiteInterpreterGetOutputTensor(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  int32_t output_index
)

Returns the tensor associated with the output index.

REQUIRES: 0 <= output_index < TfLiteInterpreterGetOutputTensorCount(tensor)

TfLiteInterpreterGetOutputTensorCount

TFL_CAPI_EXPORT int32_t TfLiteInterpreterGetOutputTensorCount(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Returns the number of output tensors associated with the model.

TfLiteInterpreterGetSignatureCount

TFL_CAPI_EXPORT int32_t TfLiteInterpreterGetSignatureCount(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

SignatureRunner APIs

You can run inference by either:

(i) (recommended) using the Interpreter to initialize SignatureRunner(s) and then only using SignatureRunner APIs.

(ii) only using Interpreter APIs.

NOTE:

  • Only use one of the above options to run inference, i.e. avoid mixing both SignatureRunner APIs and Interpreter APIs to run inference as they share the same underlying data (e.g. updating an input tensor “A” retrieved using the Interpreter APIs will update the state of the input tensor “B” retrieved using SignatureRunner APIs, if they point to the same underlying tensor in the model; as it is not possible for a user to debug this by analyzing the code, it can lead to undesirable behavior).
  • The TfLiteSignatureRunner type is conditionally thread-safe, provided that no two threads attempt to simultaneously access two TfLiteSignatureRunner instances that point to the same underlying signature, or access a TfLiteSignatureRunner and its underlying TfLiteInterpreter, unless all such simultaneous accesses are reads (rather than writes).
  • The lifetime of a TfLiteSignatureRunner object ends when [TfLiteSignatureRunnerDelete()](/lite/api_docs/c/group/c-api.html#group c api_1gac5969a8ab2c1fccdb804a379bcbc2b46) is called on it (or when the lifetime of the underlying TfLiteInterpreter ends but you should call [TfLiteSignatureRunnerDelete()](/lite/api_docs/c/group/c-api.html#group c api_1gac5969a8ab2c1fccdb804a379bcbc2b46) before that happens in order to avoid resource leaks).
  • You can only apply delegates to the interpreter (via TfLiteInterpreterOptions) and not to a signature. Returns the number of signatures defined in the model.

TfLiteInterpreterGetSignatureKey

TFL_CAPI_EXPORT const char * TfLiteInterpreterGetSignatureKey(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  int32_t signature_index
)

Returns the key of the Nth signature in the model, where N is specified as signature_index.

NOTE: The lifetime of the returned key is the same as (and depends on) the lifetime of interpreter.

TfLiteInterpreterGetSignatureRunner

TFL_CAPI_EXPORT[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)* TfLiteInterpreterGetSignatureRunner(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  const char *signature_key
)

Returns a new signature runner using the provided interpreter and signature key, or nullptr on failure.

NOTE: signature_key is a null-terminated C string that must match the key of a signature in the interpreter's model.

NOTE: The returned signature runner should be destroyed, by calling [TfLiteSignatureRunnerDelete()](/lite/api_docs/c/group/c-api.html#group c api_1gac5969a8ab2c1fccdb804a379bcbc2b46), before the interpreter is destroyed.

TfLiteInterpreterGetTensor

TFL_CAPI_EXPORT[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)* TfLiteInterpreterGetTensor(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  int index
)

Returns modifiable access to the tensor that corresponds to the specified index and is associated with the provided interpreter.

This requires the index to be between 0 and N - 1, where N is the number of tensors in the model.

Typically the tensors associated with the interpreter would be set during the interpreter initialization, through a mechanism like the InterpreterBuilder, and remain unchanged throughout the lifetime of the interpreter. However, there are some circumstances in which the pointer may not remain valid throughout the lifetime of the interpreter, because calls to AddTensors on the interpreter invalidate the returned pointer.

Note the difference between this function and TfLiteInterpreterGetInputTensor (or TfLiteInterpreterGetOutputTensor for that matter): TfLiteInterpreterGetTensor takes an index into the array of all tensors associated with the interpreter's model, whereas TfLiteInterpreterGetInputTensor takes an index into the array of input tensors.

The ownership of the tensor remains with the TFLite runtime, meaning the caller should not deallocate the pointer.

TfLiteInterpreterInputTensorIndices

TFL_CAPI_EXPORT const int * TfLiteInterpreterInputTensorIndices(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Returns a pointer to an array of input tensor indices.

The length of the array can be obtained via a call to TfLiteInterpreterGetInputTensorCount.

Typically the input tensors associated with an interpreter would be set during the initialization of the interpreter, through a mechanism like the InterpreterBuilder, and remain unchanged throughout the lifetime of the interpreter. However, there are some circumstances in which the pointer may not remain valid throughout the lifetime of the interpreter, because calls to SetInputs on the interpreter invalidate the returned pointer.

The ownership of the array remains with the TFLite runtime.

TfLiteInterpreterInvoke

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteInterpreterInvoke([TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group__c__api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Runs inference for the loaded graph.

Before calling this function, the caller should first invoke [TfLiteInterpreterAllocateTensors()](/lite/api_docs/c/group/c-api.html#group c api_1gaf9dfd1837fd95c9813cb6bb5a99ffec4) and should also set the values for the input tensors. After successfully calling this function, the values for the output tensors will be set.

If the (experimental!) delegate fallback option was enabled in the interpreter options, then the interpreter will automatically fall back to not using any delegates if execution with delegates fails. For details, see TfLiteInterpreterOptionsSetEnableDelegateFallback in [c_api_experimental.h](/lite/api_docs/c/c/api-experimental-8h-source.html#c api experimental_8h_source).

Returns one of the following status codes:

  • kTfLiteOk: Success. Output is valid.
  • kTfLiteDelegateError: Execution with delegates failed, due to a problem with the delegate(s). If fallback was not enabled, output is invalid. If fallback was enabled, this return value indicates that fallback succeeded, the output is valid, and all delegates previously applied to the interpreter have been undone.
  • kTfLiteApplicationError: Same as for kTfLiteDelegateError, except that the problem was not with the delegate itself, but rather was due to an incompatibility between the delegate(s) and the interpreter or model.
  • kTfLiteError: Unexpected/runtime failure. Output is invalid.

TfLiteInterpreterOptionsAddDelegate

TFL_CAPI_EXPORT void TfLiteInterpreterOptionsAddDelegate([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*options,[TfLiteOpaqueDelegate](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gac2bc3e65b2b4dfe997134c006faa442f)*delegate
)

Adds a delegate to be applied during TfLiteInterpreter creation.

If delegate application fails, interpreter creation will also fail with an associated error logged.

If you are NOT using "TensorFlow Lite in Play Services", and NOT building with TFLITE_WITH_STABLE_ABI or TFLITE_USE_OPAQUE_DELEGATE macros enabled, it is possible to pass a TfLiteDelegate* rather than a TfLiteOpaqueDelegate* to this function, since in those cases, TfLiteOpaqueDelegate is just a typedef alias for TfLiteDelegate. This is for compatibility with existing source code and existing delegates. For new delegates, it is recommended to use TfLiteOpaqueDelegate rather than TfLiteDelegate. (See TfLiteOpaqueDelegate in [tensorflow/lite/core/c/c_api_types.h](/lite/api_docs/c/c/api-types-8h-source.html#c api types_8h_source).)

TfLiteInterpreterOptionsAddRegistrationExternal

TFL_CAPI_EXPORT void TfLiteInterpreterOptionsAddRegistrationExternal([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*options,[TfLiteRegistrationExternal](/lite/api_docs/c/group/common.html#group__common_1gac0d70c820dfb7187a23e38e14deec7eb)*registration
)

Adds an op registration to be applied during TfLiteInterpreter creation.

The TfLiteRegistrationExternal object is needed to implement custom op of TFLite Interpreter via C API. Calling this function ensures that any TfLiteInterpreter created with the specified options can execute models that use the custom operator specified in registration. Please refer https://www.tensorflow.org/lite/guide/ops_custom for custom op support. This is an experimental API and subject to change.

TfLiteInterpreterOptionsCopy

TFL_CAPI_EXPORT[TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)* TfLiteInterpreterOptionsCopy(
  const[TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*from
)

Creates and returns a shallow copy of an options object.

The caller is responsible for calling TfLiteInterpreterOptionsDelete to deallocate the object pointed to by the returned pointer.

TfLiteInterpreterOptionsCreate

TFL_CAPI_EXPORT[TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)* TfLiteInterpreterOptionsCreate()

Returns a new interpreter options instances.

TfLiteInterpreterOptionsDelete

TFL_CAPI_EXPORT void TfLiteInterpreterOptionsDelete([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*options
)

Destroys the interpreter options instance.

TfLiteInterpreterOptionsEnableCancellation

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteInterpreterOptionsEnableCancellation([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group__c__api_1gacbeed8c564457ffba2504486b6c351f0)*options,
  bool enable
)

Enables users to cancel in-flight invocations with TfLiteInterpreterCancel.

By default it is disabled and calling to TfLiteInterpreterCancel will return kTfLiteError. See TfLiteInterpreterCancel.

TfLiteInterpreterOptionsSetErrorReporter

TFL_CAPI_EXPORT void TfLiteInterpreterOptionsSetErrorReporter([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*options,
  void(*)(void *user_data, const char *format, va_list args) reporter,
  void *user_data
)

Sets a custom error reporter for interpreter execution.

  • reporter takes the provided user_data object, as well as a C-style format string and arg list (see also vprintf).
  • user_data is optional. If non-null, it is owned by the client and must remain valid for the duration of the interpreter lifetime.

TfLiteInterpreterOptionsSetNumThreads

TFL_CAPI_EXPORT void TfLiteInterpreterOptionsSetNumThreads([TfLiteInterpreterOptions](/lite/api_docs/c/group/c-api.html#group __c__ api_1gacbeed8c564457ffba2504486b6c351f0)*options,
  int32_t num_threads
)

Sets the number of CPU threads to use for the interpreter.

TfLiteInterpreterOutputTensorIndices

TFL_CAPI_EXPORT const int * TfLiteInterpreterOutputTensorIndices(
  const[TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter
)

Returns a pointer to an array of output tensor indices.

The length of the array can be obtained via a call to TfLiteInterpreterGetOutputTensorCount.

Typically the output tensors associated with an interpreter would be set during the initialization of the interpreter, through a mechanism like the InterpreterBuilder, and remain unchanged throughout the lifetime of the interpreter. However, there are some circumstances in which the pointer may not remain valid throughout the lifetime of the interpreter, because calls to SetOutputs on the interpreter invalidate the returned pointer.

The ownership of the array remains with the TFLite runtime.

TfLiteInterpreterResizeInputTensor

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteInterpreterResizeInputTensor([TfLiteInterpreter](/lite/api_docs/c/group/c-api.html#group__c__api_1ga58688a4cf8e43f56b7796925b134f175)*interpreter,
  int32_t input_index,
  const int *input_dims,
  int32_t input_dims_size
)

Resizes the specified input tensor.

REQUIRES: 0 <= input_index < TfLiteInterpreterGetInputTensorCount(tensor)

This function makes a copy of the input dimensions, so the client can safely deallocate input_dims immediately after this function returns.

TfLiteModelCreate

TFL_CAPI_EXPORT[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)* TfLiteModelCreate(
  const void *model_data,
  size_t model_size
)

Returns a model from the provided buffer, or null on failure.

TfLiteModelCreateFromFile

TFL_CAPI_EXPORT[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)* TfLiteModelCreateFromFile(
  const char *model_path
)

Returns a model from the provided file, or null on failure.

TfLiteModelCreateFromFileWithErrorReporter

TFL_CAPI_EXPORT[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)* TfLiteModelCreateFromFileWithErrorReporter(
  const char *model_path,
  void(*)(void *user_data, const char *format, va_list args) reporter,
  void *user_data
)

Same as TfLiteModelCreateFromFile with customizble error reporter.

  • reporter takes the provided user_data object, as well as a C-style format string and arg list (see also vprintf).
  • user_data is optional. If non-null, it is owned by the client and must remain valid for the duration of the interpreter lifetime.

TfLiteModelCreateWithErrorReporter

TFL_CAPI_EXPORT[TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)* TfLiteModelCreateWithErrorReporter(
  const void *model_data,
  size_t model_size,
  void(*)(void *user_data, const char *format, va_list args) reporter,
  void *user_data
)

Same as TfLiteModelCreate with customizble error reporter.

  • reporter takes the provided user_data object, as well as a C-style format string and arg list (see also vprintf).
  • user_data is optional. If non-null, it is owned by the client and must remain valid for the duration of the interpreter lifetime.

TfLiteModelDelete

TFL_CAPI_EXPORT void TfLiteModelDelete([TfLiteModel](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga4edaf7802d29c53df72ed1410289d1be)*model
)

Destroys the model instance.

TfLiteSchemaVersion

TFL_CAPI_EXPORT int TfLiteSchemaVersion(
  void
)

The supported TensorFlow Lite model file Schema version.

Returns the (major) version number of the Schema used for model files that is supported by the (potentially dynamically loaded) TensorFlow Lite Runtime.

Model files using schema versions different to this may not be supported by the current version of the TF Lite Runtime.

TfLiteSignatureRunnerAllocateTensors

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteSignatureRunnerAllocateTensors([TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group__c__api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner
)

Updates allocations for tensors associated with a signature and resizes dependent tensors using the specified input tensor dimensionality.

This is a relatively expensive operation and hence should only be called after initializing the signature runner object and/or resizing any inputs.

TfLiteSignatureRunnerDelete

TFL_CAPI_EXPORT void TfLiteSignatureRunnerDelete([TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner
)

Destroys the signature runner.

TfLiteSignatureRunnerGetInputCount

TFL_CAPI_EXPORT size_t TfLiteSignatureRunnerGetInputCount(
  const[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner
)

Returns the number of inputs associated with a signature.

TfLiteSignatureRunnerGetInputName

TFL_CAPI_EXPORT const char * TfLiteSignatureRunnerGetInputName(
  const[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner,
  int32_t input_index
)

Returns the (null-terminated) name of the Nth input in a signature, where N is specified as input_index.

NOTE: The lifetime of the returned name is the same as (and depends on) the lifetime of signature_runner.

TfLiteSignatureRunnerGetInputTensor

TFL_CAPI_EXPORT[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)* TfLiteSignatureRunnerGetInputTensor([TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner,
  const char *input_name
)

Returns the input tensor identified by input_name in the given signature.

Returns nullptr if the given name is not valid.

NOTE: The lifetime of the returned tensor is the same as (and depends on) the lifetime of signature_runner.

TfLiteSignatureRunnerGetOutputCount

TFL_CAPI_EXPORT size_t TfLiteSignatureRunnerGetOutputCount(
  const[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner
)

Returns the number of output tensors associated with the signature.

TfLiteSignatureRunnerGetOutputName

TFL_CAPI_EXPORT const char * TfLiteSignatureRunnerGetOutputName(
  const[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner,
  int32_t output_index
)

Returns the (null-terminated) name of the Nth output in a signature, where N is specified as output_index.

NOTE: The lifetime of the returned name is the same as (and depends on) the lifetime of signature_runner.

TfLiteSignatureRunnerGetOutputTensor

TFL_CAPI_EXPORT const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)* TfLiteSignatureRunnerGetOutputTensor(
  const[TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group __c__ api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner,
  const char *output_name
)

Returns the output tensor identified by output_name in the given signature.

Returns nullptr if the given name is not valid.

NOTE: The lifetime of the returned tensor is the same as (and depends on) the lifetime of signature_runner.

TfLiteSignatureRunnerInvoke

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteSignatureRunnerInvoke([TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group__c__api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner
)

Runs inference on a given signature.

Before calling this function, the caller should first invoke [TfLiteSignatureRunnerAllocateTensors()](/lite/api_docs/c/group/c-api.html#group c api_1ga3d00a54fd91b73b351ff837e7afcd688) and should also set the values for the input tensors. After successfully calling this function, the values for the output tensors will be set.

TfLiteSignatureRunnerResizeInputTensor

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteSignatureRunnerResizeInputTensor([TfLiteSignatureRunner](/lite/api_docs/c/group/c-api.html#group__c__api_1ga0c696ec6f394f670de745b04a57e7e2c)*signature_runner,
  const char *input_name,
  const int *input_dims,
  int32_t input_dims_size
)

Resizes the input tensor identified as input_name to be the dimensions specified by input_dims and input_dims_size.

Only unknown dimensions can be resized with this function. Unknown dimensions are indicated as -1 in the dims_signature attribute of a TfLiteTensor.

Returns status of failure or success. Note that this doesn't actually resize any existing buffers. A call to [TfLiteSignatureRunnerAllocateTensors()](/lite/api_docs/c/group/c-api.html#group c api_1ga3d00a54fd91b73b351ff837e7afcd688) is required to change the tensor input buffer.

NOTE: This function is similar to TfLiteInterpreterResizeInputTensorStrict() and not [TfLiteInterpreterResizeInputTensor()](/lite/api_docs/c/group/c-api.html#group c api_1gab857f13dae3969753c697febfabe5b45).

NOTE: input_name must match the name of an input in the signature.

NOTE: This function makes a copy of the input dimensions, so the caller can safely deallocate input_dims immediately after this function returns.

TfLiteTensorByteSize

TFL_CAPI_EXPORT size_t TfLiteTensorByteSize(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns the size of the underlying data in bytes.

TfLiteTensorCopyFromBuffer

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteTensorCopyFromBuffer([TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor,
  const void *input_data,
  size_t input_data_size
)

Copies from the provided input buffer into the tensor's buffer.

REQUIRES: input_data_size == TfLiteTensorByteSize(tensor)

TfLiteTensorCopyToBuffer

TFL_CAPI_EXPORT[TfLiteStatus](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1gacf79d2fb5fa520303014d1303f1f6361)TfLiteTensorCopyToBuffer(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*output_tensor,
  void *output_data,
  size_t output_data_size
)

Copies to the provided output buffer from the tensor's buffer.

REQUIRES: output_data_size == TfLiteTensorByteSize(tensor)

TfLiteTensorData

TFL_CAPI_EXPORT void * TfLiteTensorData(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns a pointer to the underlying data buffer.

TfLiteTensorDim

TFL_CAPI_EXPORT int32_t TfLiteTensorDim(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor,
  int32_t dim_index
)

Returns the length of the tensor in the "dim_index" dimension.

REQUIRES: 0 <= dim_index < TFLiteTensorNumDims(tensor)

TfLiteTensorName

TFL_CAPI_EXPORT const char * TfLiteTensorName(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns the (null-terminated) name of the tensor.

TfLiteTensorNumDims

TFL_CAPI_EXPORT int32_t TfLiteTensorNumDims(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns the number of dimensions that the tensor has.

Returns -1 in case the 'opaque_tensor' does not have its dimensions property set.

TfLiteTensorQuantizationParams

TFL_CAPI_EXPORT[TfLiteQuantizationParams](/lite/api_docs/c/struct/tf-lite-quantization-params.html#struct_tf_lite_quantization_params)TfLiteTensorQuantizationParams(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns the parameters for asymmetric quantization.

The quantization parameters are only valid when the tensor type is kTfLiteUInt8 and the scale != 0. Quantized values can be converted back to float using: real_value = scale * (quantized_value - zero_point);

TfLiteTensorType

TFL_CAPI_EXPORT[TfLiteType](/lite/api_docs/c/group/c-api-types.html#group __c__ api__types_1ga8a47ba81bdef28b5c479ee7928a7d123)TfLiteTensorType(
  const[TfLiteTensor](/lite/api_docs/c/struct/tf-lite-tensor.html#struct_tf_lite_tensor)*tensor
)

Returns the type of a tensor element.

TfLiteVersion

TFL_CAPI_EXPORT const char * TfLiteVersion(
  void
)

The TensorFlow Lite Runtime version.

Returns a pointer to a statically allocated string that is the version number of the (potentially dynamically loaded) TF Lite Runtime library. TensorFlow Lite uses semantic versioning, and the return value should be in semver 2 format http://semver.org, starting with MAJOR.MINOR.PATCH, e.g. "2.12.0" or "2.13.0-rc2".