NeuralNetworks
#include <NeuralNetworks.h>
Summary
Typedefs |
|
---|---|
ANeuralNetworksBurst
|
typedefstruct ANeuralNetworksBurst
ANeuralNetworksBurst is an opaque type that can be used to reduce the latency of a rapid sequence of executions. |
ANeuralNetworksCompilation
|
typedefstruct ANeuralNetworksCompilation
ANeuralNetworksCompilation is an opaque type that can be used to compile a machine learning model. |
ANeuralNetworksDevice
|
typedefstruct ANeuralNetworksDevice
ANeuralNetworksDevice is an opaque type that represents a device. |
ANeuralNetworksEvent
|
typedefstruct ANeuralNetworksEvent
ANeuralNetworksEvent is an opaque type that represents an event that will be signaled once an execution completes. |
ANeuralNetworksExecution
|
typedefstruct ANeuralNetworksExecution
ANeuralNetworksExecution is an opaque type that can be used to apply a machine learning model to a set of inputs. |
ANeuralNetworksMemory
|
typedefstruct ANeuralNetworksMemory
ANeuralNetworksMemory is an opaque type that represents memory. |
ANeuralNetworksModel
|
typedefstruct ANeuralNetworksModel
ANeuralNetworksModel is an opaque type that contains a description of the mathematical operations that constitute the model. |
ANeuralNetworksOperandType
|
typedefstruct ANeuralNetworksOperandType
ANeuralNetworksOperandType describes the type of an operand. |
ANeuralNetworksOperationType
|
typedefint32_t
|
ANeuralNetworksSymmPerChannelQuantParams
|
typedef Parameters for ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL operand. |
Functions |
|
---|---|
ANeuralNetworksBurst_create(ANeuralNetworksCompilation *compilation, ANeuralNetworksBurst **burst)
|
int
Create a ANeuralNetworksBurst to apply the given compilation.
|
ANeuralNetworksBurst_free(ANeuralNetworksBurst *burst)
|
void
Destroys the burst object.
|
ANeuralNetworksCompilation_create(ANeuralNetworksModel *model, ANeuralNetworksCompilation **compilation)
|
int
Create a ANeuralNetworksCompilation to compile the given model.
|
ANeuralNetworksCompilation_createForDevices(ANeuralNetworksModel *model, const ANeuralNetworksDevice *const *devices, uint32_t numDevices, ANeuralNetworksCompilation **compilation)
|
int
Create a ANeuralNetworksCompilation to compile the given model for a specified set of devices.
|
ANeuralNetworksCompilation_finish(ANeuralNetworksCompilation *compilation)
|
int
Indicate that we have finished modifying a compilation.
|
ANeuralNetworksCompilation_free(ANeuralNetworksCompilation *compilation)
|
void
Destroy a compilation.
|
ANeuralNetworksCompilation_setCaching(ANeuralNetworksCompilation *compilation, const char *cacheDir, const uint8_t *token)
|
int
Sets the compilation caching signature and the cache directory.
|
ANeuralNetworksCompilation_setPreference(ANeuralNetworksCompilation *compilation, int32_t preference)
|
int
Sets the execution preference.
|
ANeuralNetworksDevice_getFeatureLevel(const ANeuralNetworksDevice *device, int64_t *featureLevel)
|
int
Get the supported NNAPI version of the specified device.
|
ANeuralNetworksDevice_getName(const ANeuralNetworksDevice *device, const char **name)
|
int
Get the name of the specified device.
|
ANeuralNetworksDevice_getType(const ANeuralNetworksDevice *device, int32_t *type)
|
int
Get the type of a given device.
|
ANeuralNetworksDevice_getVersion(const ANeuralNetworksDevice *device, const char **version)
|
int
Get the version of the driver implementation of the specified device.
|
ANeuralNetworksEvent_free(ANeuralNetworksEvent *event)
|
void
Destroys the event.
|
ANeuralNetworksEvent_wait(ANeuralNetworksEvent *event)
|
int
Waits until the execution completes.
|
ANeuralNetworksExecution_burstCompute(ANeuralNetworksExecution *execution, ANeuralNetworksBurst *burst)
|
int
Schedule synchronous evaluation of the execution on a burst object.
|
ANeuralNetworksExecution_compute(ANeuralNetworksExecution *execution)
|
int
Schedule synchronous evaluation of the execution.
|
ANeuralNetworksExecution_create(ANeuralNetworksCompilation *compilation, ANeuralNetworksExecution **execution)
|
int
Create a ANeuralNetworksExecution to apply the given compilation.
|
ANeuralNetworksExecution_free(ANeuralNetworksExecution *execution)
|
void
Destroy an execution.
|
ANeuralNetworksExecution_getDuration(const ANeuralNetworksExecution *execution, int32_t durationCode, uint64_t *duration)
|
int
Get the time spent in the specified ANeuralNetworksExecution, in nanoseconds.
|
ANeuralNetworksExecution_getOutputOperandDimensions(ANeuralNetworksExecution *execution, int32_t index, uint32_t *dimensions)
|
int
Get the dimensional information of the specified output operand of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_getOutputOperandRank(ANeuralNetworksExecution *execution, int32_t index, uint32_t *rank)
|
int
Get the dimensional information of the specified output operand of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_setInput(ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const void *buffer, size_t length)
|
int
Associate a user buffer with an input of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_setInputFromMemory(ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const ANeuralNetworksMemory *memory, size_t offset, size_t length)
|
int
Associate a region of a memory object with an input of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_setMeasureTiming(ANeuralNetworksExecution *execution, bool measure)
|
int
Specifies whether duration of the ANeuralNetworksExecution is to be measured.
|
ANeuralNetworksExecution_setOutput(ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, void *buffer, size_t length)
|
int
Associate a user buffer with an output of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_setOutputFromMemory(ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const ANeuralNetworksMemory *memory, size_t offset, size_t length)
|
int
Associate a region of a memory object with an output of the model of the ANeuralNetworksExecution.
|
ANeuralNetworksExecution_startCompute(ANeuralNetworksExecution *execution, ANeuralNetworksEvent **event)
|
int
Schedule asynchronous evaluation of the execution.
|
ANeuralNetworksMemory_createFromAHardwareBuffer(const AHardwareBuffer *ahwb, ANeuralNetworksMemory **memory)
|
int
Creates a shared memory object from an AHardwareBuffer handle.
|
ANeuralNetworksMemory_createFromFd(size_t size, int protect, int fd, size_t offset, ANeuralNetworksMemory **memory)
|
int
Creates a shared memory object from a file descriptor.
|
ANeuralNetworksMemory_free(ANeuralNetworksMemory *memory)
|
void
Delete a memory object.
|
ANeuralNetworksModel_addOperand(ANeuralNetworksModel *model, const ANeuralNetworksOperandType *type)
|
int
Add an operand to a model.
|
ANeuralNetworksModel_addOperation(ANeuralNetworksModel *model, ANeuralNetworksOperationType type, uint32_t inputCount, const uint32_t *inputs, uint32_t outputCount, const uint32_t *outputs)
|
int
Add an operation to a model.
|
ANeuralNetworksModel_create(ANeuralNetworksModel **model)
|
int
Create an empty ANeuralNetworksModel.
|
ANeuralNetworksModel_finish(ANeuralNetworksModel *model)
|
int
Indicate that we have finished modifying a model.
|
ANeuralNetworksModel_free(ANeuralNetworksModel *model)
|
void
Destroy a model.
|
ANeuralNetworksModel_getSupportedOperationsForDevices(const ANeuralNetworksModel *model, const ANeuralNetworksDevice *const *devices, uint32_t numDevices, bool *supportedOps)
|
int
Get the supported operations for a specified set of devices.
|
ANeuralNetworksModel_identifyInputsAndOutputs(ANeuralNetworksModel *model, uint32_t inputCount, const uint32_t *inputs, uint32_t outputCount, const uint32_t *outputs)
|
int
Specifies which operands will be the model's inputs and outputs.
|
ANeuralNetworksModel_relaxComputationFloat32toFloat16(ANeuralNetworksModel *model, bool allow)
|
int
Specifies whether ANEURALNETWORKS_TENSOR_FLOAT32 is allowed to be calculated with range and/or precision as low as that of the IEEE 754 16-bit floating-point format.
|
ANeuralNetworksModel_setOperandSymmPerChannelQuantParams(ANeuralNetworksModel *model, int32_t index, const ANeuralNetworksSymmPerChannelQuantParams *channelQuant)
|
int
Sets an operand's per channel quantization parameters.
|
ANeuralNetworksModel_setOperandValue(ANeuralNetworksModel *model, int32_t index, const void *buffer, size_t length)
|
int
Sets an operand to a constant value.
|
ANeuralNetworksModel_setOperandValueFromMemory(ANeuralNetworksModel *model, int32_t index, const ANeuralNetworksMemory *memory, size_t offset, size_t length)
|
int
Sets an operand to a value stored in a memory object.
|
ANeuralNetworks_getDevice(uint32_t devIndex, ANeuralNetworksDevice **device)
|
int
Get the representation of the specified device.
|
ANeuralNetworks_getDeviceCount(uint32_t *numDevices)
|
int
Get the number of available devices.
|
Structs |
|
---|---|
ANeuralNetworksOperandType |
ANeuralNetworksOperandType describes the type of an operand. |
ANeuralNetworksSymmPerChannelQuantParams |
Parameters for ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL operand. |
Enumerations
Anonymous Enum 45
Anonymous Enum 45
For ANeuralNetworksModel_setOperandValue, values with a length smaller or equal to this will be immediately copied into the model.
The size is in bytes.
Available since API level 27.
Properties | |
---|---|
ANEURALNETWORKS_MAX_SIZE_OF_IMMEDIATELY_COPIED_VALUES
|
Anonymous Enum 46
Anonymous Enum 46
For ANeuralNetworksCompilation_setCaching, specify the size of the cache token required from the application.
The size is in bytes.
Available since API level 29.
Properties | |
---|---|
ANEURALNETWORKS_BYTE_SIZE_OF_CACHE_TOKEN
|
DeviceTypeCode
DeviceTypeCode
Device types.
The type of NNAPI device.
DurationCode
DurationCode
FuseCode
FuseCode
Fused activation function types.
Available since API level 27.
OperandCode
OperandCode
Operand types.
The type of an operand in a model.
Types prefaced with ANEURALNETWORKS_TENSOR_* must be used for tensor data (i.e., tensors with at least one dimension). Types not prefaced by ANEURALNETWORKS_TENSOR_* represent scalar values and must have no dimensions.
Although we define many types, most operators accept just a few types. Most used are ANEURALNETWORKS_TENSOR_FLOAT32, ANEURALNETWORKS_TENSOR_QUANT8_ASYMM, and ANEURALNETWORKS_INT32.
Available since API level 27.
Properties | |
---|---|
ANEURALNETWORKS_BOOL
|
An 8 bit boolean scalar value. Values of this operand type are either true or false. A zero value represents false; any other value represents true. Available since API level 29. |
ANEURALNETWORKS_FLOAT16
|
An IEEE 754 16 bit floating point scalar value. Available since API level 29. |
ANEURALNETWORKS_FLOAT32
|
A 32 bit floating point scalar value. |
ANEURALNETWORKS_INT32
|
A signed 32 bit integer scalar value. |
ANEURALNETWORKS_TENSOR_BOOL8
|
A tensor of 8 bit boolean values. Values of this operand type are either true or false. A zero value represents false; any other value represents true. Available since API level 29. |
ANEURALNETWORKS_TENSOR_FLOAT16
|
A tensor of IEEE 754 16 bit floating point values. Available since API level 29. |
ANEURALNETWORKS_TENSOR_FLOAT32
|
A tensor of 32 bit floating point values. |
ANEURALNETWORKS_TENSOR_INT32
|
A tensor of 32 bit integer values. |
ANEURALNETWORKS_TENSOR_QUANT16_ASYMM
|
A tensor of 16 bit unsigned integers that represent real numbers. Attached to this tensor are two numbers that can be used to convert the 16 bit integer to the real value and vice versa. These two numbers are:
The formula is: real_value = (integer_value - zeroPoint) * scale. Available since API level 29. |
ANEURALNETWORKS_TENSOR_QUANT16_SYMM
|
A tensor of 16 bit signed integers that represent real numbers. Attached to this tensor is a number representing real value scale that is used to convert the 16 bit number to a real value in the following way: realValue = integerValue * scale. scale is a 32 bit floating point with value greater than zero. Available since API level 29. |
ANEURALNETWORKS_TENSOR_QUANT8_ASYMM
|
A tensor of 8 bit unsigned integers that represent real numbers. Attached to this tensor are two numbers that can be used to convert the 8 bit integer to the real value and vice versa. These two numbers are:
The formula is: real_value = (integer_value - zeroPoint) * scale. |
ANEURALNETWORKS_TENSOR_QUANT8_SYMM
|
A tensor of 8 bit signed integers that represent real numbers. Attached to this tensor is a number representing real value scale that is used to convert the 8 bit number to a real value in the following way: realValue = integerValue * scale. scale is a 32 bit floating point with value greater than zero. Available since API level 29. |
ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL
|
A tensor of 8 bit signed integers that represent real numbers. This tensor is associated with additional fields that can be used to convert the 8 bit signed integer to the real value and vice versa. These fields are:
ANeuralNetworksModel_setOperandSymmPerChannelQuantParams must be used to set the parameters for an Operand of this type. The channel dimension of this tensor must not be unknown (dimensions[channelDim] != 0). The formula is: realValue[..., C, ...] = integerValue[..., C, ...] * scales[C] where C is an index in the Channel dimension. Available since API level 29. |
ANEURALNETWORKS_UINT32
|
An unsigned 32 bit integer scalar value. |
OperationCode
OperationCode
Operation types.
The type of an operation in a model.
Available since API level 27.
Properties | |
---|---|
ANEURALNETWORKS_ABS
|
Computes the absolute value of a tensor, element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_ADD
|
Adds two tensors, element-wise. Takes two input tensors of identical OperandCode and compatible dimensions. The output is the sum of both input tensors, optionally modified by an activation function. Two dimensions are compatible when:
The size of the output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. Example: input1.dimension = {4, 1, 2} input2.dimension = {5, 4, 3, 1} output.dimension = {5, 4, 3, 2} Since API level 29, generic zero-sized input tensor is supported. Zero dimension is only compatible with 0 or 1. The size of the output dimension is zero if either of corresponding input dimension is zero. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_ARGMAX
|
Returns the index of the largest element along an axis. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_ARGMIN
|
Returns the index of the smallest element along an axis. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_AVERAGE_POOL_2D
|
Performs a 2-D average pooling operation. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, channel] = sum_{di, dj}( input[b, strides[1] * i + di, strides[2] * j + dj, channel] ) / sum(1) Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_AXIS_ALIGNED_BBOX_TRANSFORM
|
Transform axis-aligned bounding box proposals using bounding box deltas. Given the positions of bounding box proposals and the corresponding bounding box deltas for each class, return the refined bounding box regions. The resulting bounding boxes are cliped against the edges of the image. Supported tensor OperandCode: Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_BATCH_TO_SPACE_ND
|
BatchToSpace for N-dimensional tensors. This operation reshapes the batch dimension (dimension 0) into M + 1 dimensions of shape block_shape + [batch], interleaves these blocks back into the grid defined by the spatial dimensions [1, ..., M], to obtain a result with the same rank as the input. This is the reverse of SpaceToBatch. Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_LSTM
|
Performs a forward LSTM on the input followed by a backward LSTM. Supported tensor OperandCode: Supported tensor rank: 3, either time-major or batch-major. All input and output tensors must be of the same type. Inputs:
Outputs:
Available since API level 29. Important: As of API level 29, there is no way to get the output state tensors out and NNAPI does not maintain internal states. This operator does not support the usage pattern in which multiple cells are chained and state tensors are propagated. |
ANEURALNETWORKS_BIDIRECTIONAL_SEQUENCE_RNN
|
A recurrent neural network layer that applies a basic RNN cell to a sequence of inputs in forward and backward directions. This Op unrolls the input along the sequence dimension, and implements the following operation for each element in the sequence s = 1...sequence_length: fw_outputs[s] = fw_state = activation(inputs[s] * fw_input_weights’ + fw_state * fw_recurrent_weights’ + fw_bias) And for each element in sequence t = sequence_length : 1 bw_outputs[t] = bw_state = activation(inputs[t] * bw_input_weights’ + bw_state * bw_recurrent_weights’ + bw_bias) Where:
The op also supports an auxiliary input. Regular cell feeds one input into the two RNN cells in the following way: INPUT (INPUT_REVERSED) | | | FW_RNN BW_RNN || | FW_OUT BW_OUT An op with an auxiliary input takes two inputs and feeds them into the RNN cells in the following way: AUX_INPUT (AUX_INPUT_REVERSED) | | INPUT | (INPUT_R'D.)| | | | | | \ / \ / | | FW_RNN BW_RNN || | FW_OUT BW_OUT While stacking this op on top of itself, this allows to connect both forward and backward outputs from previous cell to the next cell's inputs. Supported tensor OperandCode: The input tensors must all be the same type. Inputs:
Available since API level 29. Important: As of API level 29, there is no way to get the output state tensors out and NNAPI does not maintain internal states. This operator does not support the usage pattern in which multiple cells are chained and state tensors are propagated. |
ANEURALNETWORKS_BOX_WITH_NMS_LIMIT
|
Greedily selects a subset of bounding boxes in descending order of score. This op applies NMS algorithm to each class. In each loop of execution, the box with maximum score gets selected and removed from the pending set. The scores of the rest of boxes are lowered according to the intersection-over-union (IOU) overlapping with the previously selected boxes and a specified NMS kernel method. Any boxes with score less than a threshold are removed from the pending set. Three NMS kernels are supported:
Axis-aligned bounding boxes are represented by its upper-left corner coordinate (x1,y1) and lower-right corner coordinate (x2,y2). A valid bounding box should satisfy x1 <= x2 and y1 <= y2. Supported tensor OperandCode: Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_CAST
|
Casts a tensor to a new type. This operation ignores the scale and zeroPoint of quanized tensors, e.g. it treats a ANEURALNETWORKS_TENSOR_QUANT8_ASYMM input as a tensor of uint8 values. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_CHANNEL_SHUFFLE
|
Shuffle the channels of the input tensor. Given an input tensor and a integer value of num_groups, CHANNEL_SHUFFLE divide the channel dimension into num_groups groups, and reorganize the channels by grouping channels with the same index in each group. Along the channel dimension, the output is calculated using this formula: output_channel[k * num_groups + g] = input_channel[g * group_size + k] where group_size = num_channels / num_groups The number of channels must be divisible by num_groups. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_CONCATENATION
|
Concatenates the input tensors along the given dimension. The input tensors must have identical OperandCode and the same dimensions except the dimension along the concatenation axis. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_CONV_2D
|
Performs a 2-D convolution operation. The CONV_2D op sweeps a 2-D filter that can mix channels together over a batch of images, applying the filter to each window of each image of the appropriate size. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, channel] = sum_{di, dj, k} ( input[b, strides[1] * i + di, strides[2] * j + dj, k] * filter[channel, di, dj, k] ) + bias[channel] Supported tensor OperandCode configurations:
Available since API level 29:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_DEPTHWISE_CONV_2D
|
Performs a depthwise 2-D convolution operation. Given an input tensor of shape [batches, height, width, depth_in] and a filter tensor of shape [1, filter_height, filter_width, depth_out] containing depth_out convolutional filters of depth 1, DEPTHWISE_CONV applies a different filter to each input channel (expanding from 1 channel to channel_multiplier channels for each), then concatenates the results together. The output has depth_out = depth_in * depth_multiplier channels. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, k * channel_multiplier + q] = sum_{di, dj} ( input[b, strides[1] * i + di, strides[2] * j + dj, k] * filter[1, di, dj, k * channel_multiplier + q] ) + bias[k * channel_multiplier + q] Supported tensor OperandCode configurations:
Available since API level 29:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_DEPTH_TO_SPACE
|
Rearranges data from depth into blocks of spatial data. More specifically, this op outputs a copy of the input tensor where values from the depth dimension are moved in spatial blocks to the height and width dimensions. The value block_size indicates the input block size and how the data is moved. Chunks of data of size block_size * block_size from depth are rearranged into non-overlapping blocks of size block_size x block_size. The width of the output tensor is input_depth * block_size, whereas the height is input_height * block_size. The depth of the input tensor must be divisible by block_size * block_size Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_DEQUANTIZE
|
Dequantizes the input tensor. The formula is: output = (input - zeroPoint) * scale. Supported input tensor OperandCode:
Supported output tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_DETECTION_POSTPROCESSING
|
Apply postprocessing steps to bounding box detections. Bounding box detections are generated by applying transformation on a set of predefined anchors with the bounding box deltas from bounding box regression. A final step of hard NMS is applied to limit the number of returned boxes. Supported tensor OperandCode: Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_DIV
|
Element-wise division of two tensors. Takes two input tensors of identical OperandCode and compatible dimensions. The output is the result of dividing the first input tensor by the second, optionally modified by an activation function. Two dimensions are compatible when:
The size of the output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. Example: input1.dimension = {4, 1, 2} input2.dimension = {5, 4, 3, 1} output.dimension = {5, 4, 3, 2} Since API level 29, generic zero-sized input tensor is supported. Zero dimension is only compatible with 0 or 1. The size of the output dimension is zero if either of corresponding input dimension is zero. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_EMBEDDING_LOOKUP
|
Looks up sub-tensors in the input tensor. This operator takes for input a tensor of values (Values) and a one-dimensional tensor of selection indices (Lookups). The output tensor is the concatenation of sub-tensors of Values as selected by Lookups. Think of Values as being sliced along its first dimension: The entries in Lookups select which slices are concatenated together to create the output tensor. For example, if Values has shape of [40, 200, 300] and Lookups has shape of [3], all three values found in Lookups are expected to be between 0 and 39. The resulting tensor must have shape of [3, 200, 300]. If a value in Lookups is out of bounds, the operation must fail and an error must be reported. Supported value tensor OperandCode: Supported value tensor rank: from 2 Inputs:
Output:
Available since API level 27. |
ANEURALNETWORKS_EQUAL
|
For input tensors x and y, computes x == y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_EXP
|
Computes exponential of x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_EXPAND_DIMS
|
Inserts a dimension of 1 into a tensor's shape. Given a tensor input, this operation inserts a dimension of 1 at the given dimension index of input's shape. The dimension index starts at zero; if you specify a negative dimension index, it is counted backward from the end. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_FLOOR
|
Computes element-wise floor() on the input tensor. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_FULLY_CONNECTED
|
Denotes a fully (densely) connected layer, which connects all elements in the input tensor with each element in the output tensor. This layer implements the operation: outputs = activation(inputs * weights’ + bias) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_GATHER
|
Gathers values along an axis. Produces an output tensor with shape input0.dimension[:axis] + indices.dimension + input0.dimension[axis + 1:] where: Vector indices (output is rank(input0)).output[a_0, ..., a_n, i, b_0, ..., b_n] = input0[a_0, ..., a_n, indices[i], b_0, ..., b_n]
Higher rank indices (output is rank(input0) + rank(indices) - 1).output[a_0, ..., a_n, i, ..., j, b_0, ... b_n] = input0[a_0, ..., a_n, indices[i, ..., j], b_0, ..., b_n] Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_GENERATE_PROPOSALS
|
Generate aixs-aligned bounding box proposals. Bounding box proposals are generated by applying transformation on a set of predefined anchors with the bounding box deltas from bounding box regression. A final step of hard NMS is applied to limit the number of returned boxes. Axis-aligned bounding boxes are represented by its upper-left corner coordinate (x1,y1) and lower-right corner coordinate (x2,y2). A valid bounding box should satisfy x1 <= x2 and y1 <= y2. Supported tensor OperandCode: Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_GREATER
|
For input tensors x and y, computes x > y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_GREATER_EQUAL
|
For input tensors x and y, computes x >= y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_GROUPED_CONV_2D
|
Performs a grouped 2-D convolution operation. Given an input tensor of shape [batches, height, width, depth_in] and a filter tensor of shape [depth_out, filter_height, filter_width, depth_group] containing depth_out convolutional filters of depth depth_group, GROUPED_CONV applies a group of different filters to each input channel group, then concatenates the results together. Specifically, the input channels are divided into num_groups groups, each with depth depth_group, i.e. depth_in = num_groups * depth_group. The convolutional filters are also divided into num_groups groups, i.e. depth_out is divisible by num_groups. GROUPED_CONV applies each group of filters to the corresponding input channel group, and the result are concatenated together. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, g * channel_multiplier + q] = sum_{di, dj, dk} ( input[b, strides[1] * i + di, strides[2] * j + dj, g * depth_group + dk] * filter[g * channel_multiplier + q, di, dj, dk] ) + bias[channel] where channel_multiplier = depth_out / num_groups Supported tensor OperandCode configurations:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 29. |
ANEURALNETWORKS_HASHTABLE_LOOKUP
|
Looks up sub-tensors in the input tensor using a key-value map. This operator takes for input a tensor of values (Values), a one-dimensional tensor of selection values (Lookups) and a one-dimensional tensor that maps these values to Values indexes. The output tensor is the concatenation of sub-tensors of Values as selected by Lookups via Keys. Think of Values as being sliced along its outer-most dimension. The output is a concatenation of selected slices, with one slice for each entry of Lookups. The slice selected is the one at the same index as the Maps entry that matches the value in Lookups. For a hit, the corresponding sub-tensor of Values is included in the Output tensor. For a miss, the corresponding sub-tensor in Output must have zero values. For example, if Values has shape of [40, 200, 300], Keys should have a shape of [40]. If Lookups tensor has shape of [3], three slices are being concatenated, so the resulting tensor must have the shape of [3, 200, 300]. If the first entry in Lookups has the value 123456, that value must be located in Keys tensor. If the sixth entry of Keys contains 123456, the sixth slice of Values must be selected. If no entry in Keys has 123456, a slice of zeroes must be concatenated. Supported value tensor OperandCode: Supported value tensor rank: from 2 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_HEATMAP_MAX_KEYPOINT
|
Localize the maximum keypoints from heatmaps. This operation approximates the accurate maximum keypoint scores and indices after bicubic upscaling by using Taylor expansion up to the quadratic term. The bounding box is represented by its upper-left corner coordinate (x1,y1) and lower-right corner coordinate (x2,y2) in the original image. A valid bounding box should satisfy x1 <= x2 and y1 <= y2. Supported tensor OperandCode: Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_INSTANCE_NORMALIZATION
|
Applies instance normalization to the input tensor. The values in the output tensor are computed as: output[b, h, w, c] = (input[b, h, w, c] - mean[b, c]) * gamma / sqrt(var[b, c] + epsilon) + beta Where the mean and variance are computed across the spatial dimensions: mean[b, c] = sum_{h, w}(input[b, h, w, c]) / sum(1) var[b, c] = sum_{h, w}(pow(input[b, h, w, c] - mean[b, c], 2)) / sum(1) Supported tensor OperandCode: Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_L2_NORMALIZATION
|
Applies L2 normalization along the depth dimension. The values in the output tensor are computed as: output[batch, row, col, channel] = input[batch, row, col, channel] / sqrt(sum_{c} pow(input[batch, row, col, c], 2)) For input tensor with rank less than 4, independently normalizes each 1-D slice along dimension dim. Supported tensor OperandCode:
Supported tensor rank: up to 4 Tensors with rank less than 4 are only supported since API level 29. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_L2_POOL_2D
|
Performs an 2-D L2 pooling operation. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, c] = sqrt(sum_{di, dj} pow(input[b, strides[1] * i + di, strides[2] * j + dj, c], 2) / sum(1)) Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_LESS
|
For input tensors x and y, computes x < y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LESS_EQUAL
|
For input tensors x and y, computes x <= y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LOCAL_RESPONSE_NORMALIZATION
|
Applies Local Response Normalization along the depth dimension. The 4-D input tensor is treated as a 3-D array of 1-D vectors (along the last dimension), and each vector is normalized independently. Within a given vector, each component is divided by the weighted, squared sum of inputs within depth_radius. The output is calculated using this formula: sqr_sum[a, b, c, d] = sum( pow(input[a, b, c, d - depth_radius : d + depth_radius + 1], 2)) output = input / pow((bias + alpha * sqr_sum), beta) For input tensor with rank less than 4, independently normalizes each 1-D slice along specified dimension. Supported tensor OperandCode:
Supported tensor rank: up to 4 Tensors with rank less than 4 are only supported since API level 29. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_LOG
|
Computes natural logarithm of x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LOGICAL_AND
|
Returns the truth value of x AND y element-wise. Supported tensor OperandCode: Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LOGICAL_NOT
|
Computes the truth value of NOT x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LOGICAL_OR
|
Returns the truth value of x OR y element-wise. Supported tensor OperandCode: Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LOGISTIC
|
Computes sigmoid activation on the input tensor element-wise. The output is calculated using this formula: output = 1 / (1 + exp(-input)) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_LOG_SOFTMAX
|
Computes the log softmax activations given logits. The output is calculated using this formula: output = logits * beta - log(reduce_sum(exp(logits * beta), axis)) Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_LSH_PROJECTION
|
Projects an input to a bit vector via locality senstive hashing. Supported input tensor OperandCode:
Supported input tensor rank: from 1 Inputs:
Outputs:
Available since API level 27. The offset value for sparse projections was added in API level 29. |
ANEURALNETWORKS_LSTM
|
Performs a single time step in a Long Short-Term Memory (LSTM) layer. The LSTM operation is described by the following equations. \begin{eqnarray*} i_t =& \sigma(W_{xi}x_t+W_{hi}h_{t-1}+W_{ci}C_{t-1}+b_i) & \\ f_t =& \sigma(W_{xf}x_t+W_{hf}h_{t-1}+W_{cf}C_{t-1}+b_f) & \\ C_t =& clip(f_t \odot C_{t-1} + i_t \odot g(W_{xc}x_t+W_{hc}h_{t-1}+b_c),\ t_{cell}) & \\ o_t =& \sigma(W_{xo}x_t+W_{ho}h_{t-1}+W_{co}C_t+b_o) & \\ & & \\ & clip(W_{proj}(o_t \odot g(C_t))+b_{proj},\ t_{proj}) & if\ there\ is\ a\ projection; \\ h_t =& & \\ & o_t \odot g(C_t) & otherwise. \\ \end{eqnarray*} Where:
Since API level 29 LSTM supports layer normalization. In case layer normalization is used, the inputs to internal activation functions (sigmoid and $g$) are normalized, rescaled and recentered following an approach from section 3.1 from https://arxiv.org/pdf/1607.06450.pdf The operation has the following independently optional inputs:
References: The default non-peephole non-CIFG implementation is based on: http://www.bioinf.jku.at/publications/older/2604.pdf S. Hochreiter and J. Schmidhuber. "Long Short-Term Memory". Neural Computation, 9(8):1735-1780, 1997. The peephole implementation and projection layer is based on: https://research.google.com/pubs/archive/43905.pdf Hasim Sak, Andrew Senior, and Francoise Beaufays. "Long short-term memory recurrent neural network architectures for large scale acoustic modeling." INTERSPEECH, 2014. (However, the concept of peephole optimization was introduced in work prior to this paper.) The coupling of input and forget gate (CIFG) is based on: http://arxiv.org/pdf/1503.04069.pdf Greff et al. "LSTM: A Search Space Odyssey" The layer normalization is based on: https://arxiv.org/pdf/1607.06450.pdf Jimmy Ba et al. "Layer Normalization" Supported tensor OperandCode:
All input and output tensors must be of the same type. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_MAXIMUM
|
Returns the element-wise maximum of two tensors. Supported tensor OperandCode:
Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_MAX_POOL_2D
|
Performs an 2-D max pooling operation. The output dimensions are functions of the filter dimensions, stride, and padding. The values in the output tensor are computed as: output[b, i, j, channel] = max_{di, dj} ( input[b, strides[1] * i + di, strides[2] * j + dj, channel] ) Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_MEAN
|
Computes the mean of elements across dimensions of a tensor. Reduces the input tensor along the given dimensions to reduce. Unless keep_dims is true, the rank of the tensor is reduced by 1 for each entry in axis. If keep_dims is true, the reduced dimensions are retained with length 1. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_MINIMUM
|
Returns the element-wise minimum of two tensors. Supported tensor OperandCode:
Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_MUL
|
Multiplies two tensors, element-wise. Takes two input tensors of identical OperandCode and compatible dimensions. The output is the product of both input tensors, optionally modified by an activation function. Two dimensions are compatible when:
The size of the resulting output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. Since API level 29, generic zero-sized input tensor is supported. Zero dimension is only compatible with 0 or 1. The size of the output dimension is zero if either of corresponding input dimension is zero. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_NEG
|
Computes numerical negative value element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_NOT_EQUAL
|
For input tensors x and y, computes x != y elementwise. Supported tensor OperandCode:
Supported tensor rank: from 1 This operation supports broadcasting. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_PAD
|
Pads a tensor. This operation pads a tensor according to the specified paddings. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_PAD_V2
|
Pads a tensor with the given constant value according to the specified paddings. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_POW
|
Computes the power of one value to another. Given a tensor base and a tensor exponent, this operation computes base^exponent elementwise. This operations supports broadcasting. The size of the output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. For example: base.dimension = {4, 1, 2} exponent.dimension = {5, 4, 3, 1} output.dimension = {5, 4, 3, 2} Supported tensor OperandCode: Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_PRELU
|
Parametric Rectified Linear Unit. It follows: f(x) = alpha * x for x < 0, f(x) = x for x >= 0, where alpha is a learned array with the same OperandCode and compatible dimensions as input x. Two dimensions are compatible when:
The size of the output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. Example: input.dimension = {4, 1, 2} alpha.dimension = {5, 4, 3, 1} output.dimension = {5, 4, 3, 2} Supported tensor OperandCode: Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_QUANTIZE
|
Quantizes the input tensor. The formula is: output = max(0, min(255, round(input / scale) + zeroPoint) Supported tensor OperandCode: Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_QUANTIZED_16BIT_LSTM
|
A version of quantized LSTM, using 16 bit quantization for internal state. There is no projection layer, so cell state size is equal to the output size. Inputs:
Outputs:
|
ANEURALNETWORKS_RANDOM_MULTINOMIAL
|
Draws samples from a multinomial distribution. Supported tensor OperandCode: Inputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_ALL
|
Reduces a tensor by computing the "logical and" of elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_ANY
|
Reduces a tensor by computing the "logical or" of elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_MAX
|
Reduces a tensor by computing the maximum of elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_MIN
|
Reduces a tensor by computing the minimum of elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_PROD
|
Reduces a tensor by multiplying elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_REDUCE_SUM
|
Reduces a tensor by summing elements along given dimensions. If keep_dims is true, the reduced dimensions are retained with length 1. Otherwise, the rank of the tensor is reduced by 1 for each entry in dimensions. Supported tensor OperandCode: Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_RELU
|
Computes rectified linear activation on the input tensor element-wise. The output is calculated using this formula: output = max(0, input) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_RELU1
|
Computes rectified linear 1 activation on the input tensor element-wise. The output is calculated using this formula: output = min(1.f, max(-1.f, input)) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_RELU6
|
Computes rectified linear 6 activation on the input tensor element-wise. The output is calculated using this formula: output = min(6, max(0, input)) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_RESHAPE
|
Reshapes a tensor. Given tensor, this operation returns a tensor that has the same values as tensor, but with a newly specified shape. Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_RESIZE_BILINEAR
|
Resizes images to given size using the bilinear interpretation. Resized images must be distorted if their output aspect ratio is not the same as input aspect ratio. The corner pixels of output may not be the same as corner pixels of input. Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Both resizing by shape and resizing by scale are supported. Inputs (resizing by shape):
Inputs (resizing by scale, since API level 29):
Outputs:
Available since API level 27. |
ANEURALNETWORKS_RESIZE_NEAREST_NEIGHBOR
|
Resizes images to given size using the nearest neighbor interpretation. Resized images must be distorted if their output aspect ratio is not the same as input aspect ratio. The corner pixels of output may not be the same as corner pixels of input. Supported tensor OperandCode: Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Both resizing by shape and resizing by scale are supported. Inputs (resizing by shape):
Inputs (resizing by scale):
Outputs:
Available since API level 29. |
ANEURALNETWORKS_RNN
|
A basic recurrent neural network layer. This layer implements the operation: outputs = state = activation(inputs * input_weights + state * recurrent_weights + bias) Where:
Supported tensor OperandCode:
The input tensors must all be the same type. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_ROI_ALIGN
|
Select and scale the feature map of each region of interest to a unified output size by average pooling sampling points from bilinear interpolation. The region of interest is represented by its upper-left corner coordinate (x1,y1) and lower-right corner coordinate (x2,y2) in the original image. A spatial scaling factor is applied to map into feature map coordinate. A valid region of interest should satisfy x1 <= x2 and y1 <= y2. No rounding is applied in this operation. The sampling points are unified distributed in the pooling bin and their values are calculated by bilinear interpolation. Supported tensor OperandCode: Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_ROI_POOLING
|
Select and scale the feature map of each region of interest to a unified output size by max-pooling. The region of interest is represented by its upper-left corner coordinate (x1,y1) and lower-right corner coordinate (x2,y2) in the original image. A spatial scaling factor is applied to map into feature map coordinate. A valid region of interest should satisfy x1 <= x2 and y1 <= y2. Rounding is applied in this operation to ensure integer boundary for regions of interest and pooling bins. Supported tensor OperandCode: Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_RSQRT
|
Computes reciprocal of square root of x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_SELECT
|
Using a tensor of booleans c and input tensors x and y select values elementwise from both input tensors: O[i] = C[i] ? x[i] : y[i]. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
|
ANEURALNETWORKS_SIN
|
Computes sin of x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_SLICE
|
Extracts a slice of specified size from the input tensor starting at a specified location. The starting location is specified as a 1-D tensor containing offsets for each dimension. The size is specified as a 1-D tensor containing either size of a slice along corresponding dimension or -1. In the latter case, all the remaining elements in dimension are included in the slice. A sum of begin offset and a size of a slice must not exceed size of a corresponding dimension. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_SOFTMAX
|
Computes the softmax activation on the input tensor element-wise, per batch, by normalizing the input vector so the maximum coefficient is zero. The output is calculated using this formula: output[batch, i] = exp((input[batch, i] - max(input[batch, :])) * beta) / sum_{k}{exp((input[batch, k] - max(input[batch, :])) * beta)} For input tensor with rank other than 2, the activation will be applied independently on each 1-D slice along specified dimension. Supported tensor OperandCode:
Supported tensor rank: up to 4. Tensors with rank other than 2 or 4 are only supported since API level 29. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_SPACE_TO_BATCH_ND
|
SpaceToBatch for N-Dimensional tensors. This operation divides "spatial" dimensions [1, ..., M] of the input into a grid of blocks of shape block_shape, and interleaves these blocks with the "batch" dimension (0) such that in the output, the spatial dimensions [1, ..., M] correspond to the position within the grid, and the batch dimension combines both the position within a spatial block and the original batch position. Prior to division into blocks, the spatial dimensions of the input are optionally zero padded according to paddings. Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_SPACE_TO_DEPTH
|
Rearranges blocks of spatial data, into depth. More specifically, this op outputs a copy of the input tensor where values from the height and width dimensions are moved to the depth dimension. The value block_size indicates the input block size and how the data is moved. Chunks of data of size block_size * block_size from depth are rearranged into non-overlapping blocks of size block_size x block_size. The depth of the output tensor is input_depth * block_size * block_size. The input tensor's height and width must be divisible by block_size. Supported tensor OperandCode:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. NCHW is supported since API level 29. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_SPLIT
|
Splits a tensor along a given axis into num_splits subtensors. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_SQRT
|
Computes square root of x element-wise. Supported tensor OperandCode: Supported tensor rank: from 1. Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_SQUEEZE
|
Removes dimensions of size 1 from the shape of a tensor. Given a tensor input, this operation returns a tensor of the same OperandCode with all dimensions of size 1 removed. If you don't want to remove all size 1 dimensions, you can remove specific size 1 dimensions by specifying the axes (input1). Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_STRIDED_SLICE
|
Extracts a strided slice of a tensor. Roughly speaking, this op extracts a slice of size (end - begin) / stride from the given input tensor. Starting at the location specified by begin the slice continues by adding stride to the index until all dimensions are not less than end. Note that a stride can be negative, which causes a reverse slice. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_SUB
|
Element-wise subtraction of two tensors. Takes two input tensors of identical OperandCode and compatible dimensions. The output is the result of subtracting the second input tensor from the first one, optionally modified by an activation function. Two dimensions are compatible when:
The size of the output is the maximum size along each dimension of the input operands. It starts with the trailing dimensions, and works its way forward. Example: input1.dimension = {4, 1, 2} input2.dimension = {5, 4, 3, 1} output.dimension = {5, 4, 3, 2} Since API level 29, generic zero-sized input tensor is supported. Zero dimension is only compatible with 0 or 1. The size of the output dimension is zero if either of corresponding input dimension is zero. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_SVDF
|
SVDF op is a kind of stateful layer derived from the notion that a densely connected layer that's processing a sequence of input frames can be approximated by using a singular value decomposition of each of its nodes. The implementation is based on: https://research.google.com/pubs/archive/43813.pdf P. Nakkiran, R. Alvarez, R. Prabhavalkar, C. Parada. “Compressing Deep Neural Networks using a Rank-Constrained Topology”. INTERSPEECH, 2015. It processes the incoming input using a 2-stage filtering mechanism:
Specifically, for rank 1, this layer implements the operation: memory = push(conv1d(inputs, weights_feature, feature_dim, "ANEURALNETWORKS_PADDING_VALID")); outputs = activation(memory * weights_time + bias); Where:
Each rank adds a dimension to the weights matrices by means of stacking the filters. Supported tensor OperandCode:
All input tensors must be the same type. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_TANH
|
Computes hyperbolic tangent of input tensor element-wise. The output is calculated using this formula: output = tanh(input) Supported tensor OperandCode:
Supported tensor rank: up to 4. Inputs:
Outputs:
Available since API level 27. |
ANEURALNETWORKS_TILE
|
Constructs a tensor by tiling a given tensor. This operation creates a new tensor by replicating Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_TOPK_V2
|
Finds values and indices of the k largest entries for the last dimension. Resulting values in each dimensions are sorted in descending order. If two values are equal, the one with larger index appears first. Supported tensor OperandCode:
Supported tensor rank: from 1 Inputs:
Outputs:
Available since API level 29. |
ANEURALNETWORKS_TRANSPOSE
|
Transposes the input tensor, permuting the dimensions according to the perm tensor. The returned tensor's dimension i corresponds to the input dimension perm[i]. If perm is not given, it is set to (n-1...0), where n is the rank of the input tensor. Hence by default, this operation performs a regular matrix transpose on 2-D input Tensors. Supported tensor OperandCode:
Supported tensor rank: up to 4 Inputs:
Outputs:
Available since API level 28. |
ANEURALNETWORKS_TRANSPOSE_CONV_2D
|
Performs the transpose of 2-D convolution operation. This operation is sometimes called "deconvolution" after Deconvolutional Networks, but is actually the transpose (gradient) of ANEURALNETWORKS_CONV_2D rather than an actual deconvolution. The output dimensions are functions of the filter dimensions, stride, and padding. Supported tensor OperandCode configurations:
Supported tensor rank: 4, with "NHWC" or "NCHW" data layout. With the default data layout NHWC, the data is stored in the order of: [batch, height, width, channels]. Alternatively, the data layout could be NCHW, the data storage order of: [batch, channels, height, width]. Both explicit padding and implicit padding are supported. Inputs (explicit padding):
Inputs (implicit padding):
Outputs:
Available since API level 29. |
ANEURALNETWORKS_UNIDIRECTIONAL_SEQUENCE_LSTM
|
A recurrent neural network specified by an LSTM cell. Performs (fully) dynamic unrolling of input. This Op unrolls the input along the time dimension, and implements the following operation for each element in the sequence s = 1...sequence_length: outputs[s] = projection(state = activation(LSTMOp(inputs[s]))) Where LSTMOp is the LSTM op as in ANEURALNETWORKS_LSTM, the "projection" is an optional projection layer from state and output and the “activation” is the function passed as the “fused_activation_function” argument (if not “NONE”). Supported tensor OperandCode: Supported tensor rank: 3, either time-major or batch-major. All input and output tensors must be of the same type. Inputs:
Outputs:
Available since API level 29. Important: As of API level 29, there is no way to get the output state tensors out and NNAPI does not maintain internal states. This operator does not support the usage pattern in which multiple cells are chained and state tensors are propagated. |
ANEURALNETWORKS_UNIDIRECTIONAL_SEQUENCE_RNN
|
A recurrent neural network layer that applies a basic RNN cell to a sequence of inputs. This layer unrolls the input along the sequence dimension, and implements the following operation for each element in the sequence s = 1...sequence_length: outputs[s] = state = activation(inputs[s] * input_weights’ + state * recurrent_weights’ + bias) Where:
Supported tensor OperandCode: The input tensors must all be the same type. Inputs:
Available since API level 29. Important: As of API level 29, there is no way to get the output state tensors out and NNAPI does not maintain internal states. This operator does not support the usage pattern in which multiple cells are chained and state tensors are propagated. |
PaddingCode
PaddingCode
Implicit padding algorithms.
Available since API level 27.
PreferenceCode
PreferenceCode
Execution preferences.
Available since API level 27.
ResultCode
ResultCode
Result codes.
Any NNAPI function can return any result code, including result codes not currently documented. Any value other than ANEURALNETWORKS_NO_ERROR indicates a failure of some kind.
Additional information about the nature of a failure can be obtained from the device log after enabling NNAPI debugging by setting the debug.nn.vlog property to 1, e.g., by calling "adb shell setprop debug.nn.vlog 1".
Available since API level 27.
Typedefs
ANeuralNetworksBurst
struct ANeuralNetworksBurst ANeuralNetworksBurst
ANeuralNetworksBurst is an opaque type that can be used to reduce the latency of a rapid sequence of executions.
It will likely cause overhead if only used for a single execution.
ANeuralNetworksBurst serves as a context object for any number of inferences using ANeuralNetworksExecution objects. An ANeuralNetworksBurst object and the ANeuralNetworksExecution objects used with it must all have been created from the same ANeuralNetworksCompilation object.
This object is also used as a hint to drivers, providing insight to the lifetime of a rapid sequence of executions. For example, a driver may choose to increase the clock frequency of its accelerator for the lifetime of a burst object.
To use:
- Create a new burst object by calling the ANeuralNetworksBurst_create function.
- For each execution:
- Create ANeuralNetworksExecution and configure its properties (see ANeuralNetworksExecution for details).
- Apply the model synchronously with ANeuralNetworksExecution_burstCompute, reusing the same ANeuralNetworksBurst with the new ANeuralNetworksExecution.
- Use and free the ANeuralNetworksExecution.
- Destroy the burst with ANeuralNetworksBurst_free.
Available since API level 29.
ANeuralNetworksCompilation
struct ANeuralNetworksCompilation ANeuralNetworksCompilation
ANeuralNetworksCompilation is an opaque type that can be used to compile a machine learning model.
To use:
- Create a new compilation instance by calling the ANeuralNetworksCompilation_create function or ANeuralNetworksCompilation_createForDevices.
- Set any desired properties on the compilation (for example, ANeuralNetworksCompilation_setPreference).
- Optionally, set the caching signature and the cache directory on the compilation by calling ANeuralNetworksCompilation_setCaching.
- Complete the compilation with ANeuralNetworksCompilation_finish.
- Use the compilation as many times as needed with ANeuralNetworksExecution_create and ANeuralNetworksBurst_create.
- Destroy the compilation with ANeuralNetworksCompilation_free once all executions using the compilation have completed.
A compilation is completed by calling ANeuralNetworksCompilation_finish. A compilation is destroyed by calling ANeuralNetworksCompilation_free.
A compilation cannot be modified once ANeuralNetworksCompilation_finish has been called on it.
It is the application's responsibility to make sure that only one thread modifies a compilation at a given time. It is however safe for more than one thread to use the compilation once ANeuralNetworksCompilation_finish has returned.
It is also the application's responsibility to ensure that there are no other uses of the compilation after calling ANeuralNetworksCompilation_free. This includes any execution object or burst object created using the compilation.
Available since API level 27.
ANeuralNetworksDevice
struct ANeuralNetworksDevice ANeuralNetworksDevice
ANeuralNetworksDevice is an opaque type that represents a device.
This type is used to query basic properties and supported operations of the corresponding device, and control which device(s) a model is to be run on.
Available since API level 29.
ANeuralNetworksEvent
struct ANeuralNetworksEvent ANeuralNetworksEvent
ANeuralNetworksEvent is an opaque type that represents an event that will be signaled once an execution completes.
Available since API level 27.
ANeuralNetworksExecution
struct ANeuralNetworksExecution ANeuralNetworksExecution
ANeuralNetworksExecution is an opaque type that can be used to apply a machine learning model to a set of inputs.
To use:
- Create a new execution instance by calling the ANeuralNetworksExecution_create function.
- Associate input buffers or memory regions to the model inputs with ANeuralNetworksExecution_setInput or ANeuralNetworksExecution_setInputFromMemory.
- Associate output buffers or memory regions to the model outputs with ANeuralNetworksExecution_setOutput or ANeuralNetworksExecution_setOutputFromMemory.
- Apply the model with one of the following:
- Asynchronously with ANeuralNetworksExecution_startCompute, waiting for the execution to complete with ANeuralNetworksEvent_wait.
- Synchronously with ANeuralNetworksExecution_compute.
- Synchronously as part of an execution burst with ANeuralNetworksExecution_burstCompute.
- Destroy the execution with ANeuralNetworksExecution_free.
An output buffer or memory region must not overlap with any other output buffer or memory region, with an input buffer or memory region, or with an operand value in a memory object (ANeuralNetworksModel_setOperandValueFromMemory).
An execution cannot be modified once ANeuralNetworksExecution_burstCompute, ANeuralNetworksExecution_compute or ANeuralNetworksExecution_startCompute has been called on it.
An execution can be applied to a model with ANeuralNetworksExecution_burstCompute, ANeuralNetworksExecution_compute or ANeuralNetworksExecution_startCompute only once. Create new executions to do new evaluations of the model.
It is the application's responsibility to make sure that only one thread modifies an execution at a given time. It is however safe for more than one thread to use ANeuralNetworksEvent_wait at the same time.
It is also the application's responsibility to ensure that the execution either has never been scheduled or has completed (i.e., that ANeuralNetworksExecution_burstCompute, ANeuralNetworksExecution_compute, or ANeuralNetworksEvent_wait has returned) before calling ANeuralNetworksExecution_free.
.
It is also the application's responsibility to ensure that there are no other uses of the execution after calling ANeuralNetworksExecution_free.
Multiple executions can be scheduled and evaluated concurrently, either by means of ANeuralNetworksExecution_compute or ANeuralNetworksExecution_burstCompute (which are synchronous) in different threads, or by means of ANeuralNetworksExecution_startCompute (which is asynchronous). (Concurrent uses of ANeuralNetworksExecution_burstCompute must be on different burst objects.) The runtime makes no guarantee on the ordering of completion of executions. If it's important to the application, the application should enforce the ordering by ensuring that one execution completes before the next is scheduled (for example, by scheduling all executions synchronously within a single thread, or by scheduling all executions asynchronously and using ANeuralNetworksEvent_wait between calls to ANeuralNetworksExecution_startCompute).
Available since API level 27.
ANeuralNetworksMemory
struct ANeuralNetworksMemory ANeuralNetworksMemory
ANeuralNetworksMemory is an opaque type that represents memory.
This type is used to represent shared memory, memory mapped files, and similar memories.
By using shared memory, a program can efficiently communicate to the runtime and drivers the tensors that define a model. See ANeuralNetworksModel_setOperandValueFromMemory. An application should typically create one shared memory object that contains every constant tensor needed to define a model. ANeuralNetworksMemory_createFromFd can be used to create shared memory from a file handle. ANeuralNetworksMemory_createFromAHardwareBuffer can be used to create shared memory from an AHardwareBuffer handle.
Memory objects can also be used to specify the input and output arguments of an execution. See ANeuralNetworksExecution_setInputFromMemory and ANeuralNetworksExecution_setOutputFromMemory.
When calling ANeuralNetworksModel_setOperandValueFromMemory, ANeuralNetworksExecution_setInputFromMemory and ANeuralNetworksExecution_setOutputFromMemory, each operand in the shared memory object must be aligned on a boundary of a byte size that is a multiple of the element type byte size, e.g., a tensor with ANEURALNETWORKS_TENSOR_FLOAT32 type must be aligned on 4-byte boundary.
It is the application's responsibility to ensure that there are no uses of the memory after calling ANeuralNetworksMemory_free. This includes any model which references this memory because of a call to ANeuralNetworksModel_setOperandValueFromMemory, any compilation created using such a model, any execution object or burst object created using such a compilation, or any execution which references this memory because of a call to ANeuralNetworksExecution_setInputFromMemory or ANeuralNetworksExecution_setOutputFromMemory.
Available since API level 27.
ANeuralNetworksModel
struct ANeuralNetworksModel ANeuralNetworksModel
ANeuralNetworksModel is an opaque type that contains a description of the mathematical operations that constitute the model.
Build the model by calling
This forms a graph in which each operation and operand is a node, a directed edge from an operand to an operation indicates that the operand is an input to the operation, and a directed edge from an operation to an operand indicates that the operand is an output from the operation. This graph must be acyclic.
A model is completed by calling ANeuralNetworksModel_finish. A model is destroyed by calling ANeuralNetworksModel_free.
A model cannot be modified once ANeuralNetworksModel_finish has been called on it.
It is the application's responsibility to make sure that only one thread modifies a model at a given time. It is however safe for more than one thread to use the model once ANeuralNetworksModel_finish has returned.
It is also the application's responsibility to ensure that there are no other uses of the model after calling ANeuralNetworksModel_free. This includes any compilation, execution object or burst object created using the model.
Available since API level 27.
ANeuralNetworksOperandType
struct ANeuralNetworksOperandType ANeuralNetworksOperandType
ANeuralNetworksOperandType describes the type of an operand.
This structure is used to describe both scalars and tensors.
A tensor operand type with all dimensions specified is "fully specified". Whenever possible (i.e., whenever the dimensions are known at model construction time), a tensor operand type should be (but is not required to be) fully specified, in order to enable the best possible performance.
If a tensor operand's type is not fully specified, the dimensions of the operand are deduced from the operand types and values of the operation for which that operand is an output.
In the following situations, a tensor operand type must be fully specified:
- The operand has a constant value, set by ANeuralNetworksModel_setOperandValue (with a non-nullptr buffer) or ANeuralNetworksModel_setOperandValueFromMemory.
- The operand is a model input (see ANeuralNetworksModel_identifyInputsAndOutputs). A fully specified tensor operand type must either be provided to ANeuralNetworksModel_addOperand; or it must be provided to the corresponding ANeuralNetworksExecution_setInput, or ANeuralNetworksExecution_setInputFromMemory. EXCEPTION: If the input is optional and omitted (by passing nullptr for buffer to ANeuralNetworksExecution_setInput) then it need not have a fully specified tensor operand type.
A tensor operand type of specified rank but some number of unspecified dimensions is represented by setting dimensionCount to the rank and each unspecified dimension to 0.
Available since API level 27.
Starting at API level 29, a tensor operand type of unspecified rank is represented by setting dimensionCount to 0 and dimensions to NULL (just as if it were a scalar operand type).
ANeuralNetworksOperationType
int32_t ANeuralNetworksOperationType
ANeuralNetworksSymmPerChannelQuantParams
struct ANeuralNetworksSymmPerChannelQuantParams ANeuralNetworksSymmPerChannelQuantParams
Parameters for ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL operand.
Functions
ANeuralNetworksBurst_create
int ANeuralNetworksBurst_create( ANeuralNetworksCompilation *compilation, ANeuralNetworksBurst **burst )
Create a ANeuralNetworksBurst to apply the given compilation.
This only creates the burst object. Computation is only performed once ANeuralNetworksExecution_burstCompute is invoked with a valid ANeuralNetworksExecution and ANeuralNetworksBurst.
The provided compilation must outlive the burst object.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the compilation is invalid.
|
ANeuralNetworksBurst_free
void ANeuralNetworksBurst_free( ANeuralNetworksBurst *burst )
Destroys the burst object.
Available since API level 29.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksCompilation_create
int ANeuralNetworksCompilation_create( ANeuralNetworksModel *model, ANeuralNetworksCompilation **compilation )
Create a ANeuralNetworksCompilation to compile the given model.
This only creates the object. Compilation is only performed once ANeuralNetworksCompilation_finish is invoked.
ANeuralNetworksCompilation_finish should be called once all desired properties have been set on the compilation.
ANeuralNetworksModel_free should be called once the compilation is no longer needed.
The provided model must outlive the compilation.
The model must already have been finished by a call to ANeuralNetworksModel_finish.
See ANeuralNetworksCompilation for information on multithreaded usage.
Available since API level 27.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the model is invalid.
|
ANeuralNetworksCompilation_createForDevices
int ANeuralNetworksCompilation_createForDevices( ANeuralNetworksModel *model, const ANeuralNetworksDevice *const *devices, uint32_t numDevices, ANeuralNetworksCompilation **compilation )
Create a ANeuralNetworksCompilation to compile the given model for a specified set of devices.
If more than one device is specified, the compilation will distribute the workload automatically across the devices. The model must be fully supported by the specified set of devices. This means that ANeuralNetworksModel_getSupportedOperationsForDevices() must have returned true for every operation for that model/devices pair.
The user must handle all compilation and execution failures from the specified set of devices. This is in contrast to a use of ANeuralNetworksCompilation_create, where the runtime will attempt to recover from such failures.
Available since API level 29.
Details | |||||||||
---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the model is invalid.
|
ANeuralNetworksCompilation_finish
int ANeuralNetworksCompilation_finish( ANeuralNetworksCompilation *compilation )
Indicate that we have finished modifying a compilation.
Required before calling ANeuralNetworksExecution_create.
An application must ensure that no other thread uses the compilation at the same time.
This function must only be called once for a given compilation.
See ANeuralNetworksCompilation for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksCompilation_free
void ANeuralNetworksCompilation_free( ANeuralNetworksCompilation *compilation )
Destroy a compilation.
The compilation need not have been finished by a call to ANeuralNetworksCompilation_finish.
See ANeuralNetworksCompilation for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksCompilation_setCaching
int ANeuralNetworksCompilation_setCaching( ANeuralNetworksCompilation *compilation, const char *cacheDir, const uint8_t *token )
Sets the compilation caching signature and the cache directory.
Provides optional caching information to the runtime for faster repeated compilation.
See ANeuralNetworksCompilation for information on multithreaded usage.
Available since API level 29.
Details | |||||||
---|---|---|---|---|---|---|---|
Parameters |
|
||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksCompilation_setPreference
int ANeuralNetworksCompilation_setPreference( ANeuralNetworksCompilation *compilation, int32_t preference )
Sets the execution preference.
Provides guidance to the runtime when trade-offs are possible. By default the runtime uses PREFER_SINGLE_FAST_ANSWER
See ANeuralNetworksCompilation for information on multithreaded usage.
Available since API level 27.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksDevice_getFeatureLevel
int ANeuralNetworksDevice_getFeatureLevel( const ANeuralNetworksDevice *device, int64_t *featureLevel )
Get the supported NNAPI version of the specified device.
Each device has a supported feature level, which is the most advanced feature this driver implements. For example, if the driver implements the features introduced in Android P, but does not implement the features introduced after Android P, the value would be 28. Developers could decide whether or not the specified device should be used for a Model that has certain feature requirements.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksDevice_getName
int ANeuralNetworksDevice_getName( const ANeuralNetworksDevice *device, const char **name )
Get the name of the specified device.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksDevice_getType
int ANeuralNetworksDevice_getType( const ANeuralNetworksDevice *device, int32_t *type )
Get the type of a given device.
The device type can be used to help application developers to distribute Machine Learning workloads and other workloads such as graphical rendering. E.g., for an app which renders AR scenes based on real time object detection results, the developer could choose an ACCELERATOR type device for ML workloads, and reserve GPU for graphical rendering.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksDevice_getVersion
int ANeuralNetworksDevice_getVersion( const ANeuralNetworksDevice *device, const char **version )
Get the version of the driver implementation of the specified device.
It’s the responsibility of the driver implementor to insure that this version string uniquely distinguishes this implementation from all previous implementations.
This version string must not be confused with the feature level which is solely defined by ANeuralNetworksDevice_getFeatureLevel. There is no implicit ordering of the versions. For example, it is not possible to filter all drivers older than a certain version.
Application developers may use this version string to avoid or prefer specific driver implementations. For example, an application may want to do so because:
- A specific version of the driver does not provide the required performance, perhaps because of a performance regression.
- A specific version of the driver has a bug or returns results that don’t match the minimum precision requirement for the application.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksEvent_free
void ANeuralNetworksEvent_free( ANeuralNetworksEvent *event )
Destroys the event.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksEvent_wait
int ANeuralNetworksEvent_wait( ANeuralNetworksEvent *event )
Waits until the execution completes.
More than one thread can wait on an event. When the execution completes, all threads will be released.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |
---|---|
Returns |
ANEURALNETWORKS_NO_ERROR if the execution completed normally. ANEURALNETWORKS_UNMAPPABLE if the execution input or output memory cannot be properly mapped.
|
ANeuralNetworksExecution_burstCompute
int ANeuralNetworksExecution_burstCompute( ANeuralNetworksExecution *execution, ANeuralNetworksBurst *burst )
Schedule synchronous evaluation of the execution on a burst object.
Schedules synchronous evaluation of the execution. Returns once the execution has completed and the outputs are ready to be consumed.
There must be at most one ANeuralNetworksExecution processing at any given time for any given burst object. Any ANeuralNetworksExecution launched before the previous has finished will result in ANEURALNETWORKS_BAD_STATE.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if the execution completed normally.
|
ANeuralNetworksExecution_compute
int ANeuralNetworksExecution_compute( ANeuralNetworksExecution *execution )
Schedule synchronous evaluation of the execution.
Schedules synchronous evaluation of the execution. Returns once the execution has completed and the outputs are ready to be consumed.
See ANeuralNetworksExecution for information on multithreaded usage.
See ANeuralNetworksExecution_startCompute for asynchronous execution. Synchronous execution incurs lower overhead than asynchronous execution.
Available since API level 29.
Details | |||
---|---|---|---|
Parameters |
|
||
Returns |
ANEURALNETWORKS_NO_ERROR if the execution completed normally. ANEURALNETWORKS_UNMAPPABLE if the execution input or output memory cannot be properly mapped.
|
ANeuralNetworksExecution_create
int ANeuralNetworksExecution_create( ANeuralNetworksCompilation *compilation, ANeuralNetworksExecution **execution )
Create a ANeuralNetworksExecution to apply the given compilation.
This only creates the object. Computation is only performed once ANeuralNetworksExecution_compute or ANeuralNetworksExecution_startCompute is invoked.
The provided compilation must outlive the execution.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the compilation is invalid.
|
ANeuralNetworksExecution_free
void ANeuralNetworksExecution_free( ANeuralNetworksExecution *execution )
Destroy an execution.
The execution need not have been scheduled by a call to ANeuralNetworksExecution_burstCompute, ANeuralNetworksExecution_compute, or ANeuralNetworksExecution_startCompute; but if it has been scheduled, then the application must not call ANeuralNetworksExecution_free until the execution has completed (i.e., ANeuralNetworksExecution_burstCompute, ANeuralNetworksExecution_compute, or ANeuralNetworksEvent_wait has returned).
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksExecution_getDuration
int ANeuralNetworksExecution_getDuration( const ANeuralNetworksExecution *execution, int32_t durationCode, uint64_t *duration )
Get the time spent in the specified ANeuralNetworksExecution, in nanoseconds.
The execution must have completed.
Available since API level 29.
Details | |||||||
---|---|---|---|---|---|---|---|
Parameters |
|
||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksExecution_getOutputOperandDimensions
int ANeuralNetworksExecution_getOutputOperandDimensions( ANeuralNetworksExecution *execution, int32_t index, uint32_t *dimensions )
Get the dimensional information of the specified output operand of the model of the ANeuralNetworksExecution.
The target output operand cannot be a scalar.
On asynchronous execution initiated by ANeuralNetworksExecution_startCompute, ANeuralNetworksEvent_wait must be called prior to this function to recuperate the resources used by the execution.
Available since API level 29.
Details | |||||||
---|---|---|---|---|---|---|---|
Parameters |
|
||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_OUTPUT_INSUFFICIENT_SIZE if the target output is provided an insufficient buffer at execution time, ANEURALNETWORKS_BAD_DATA if the index is invalid or if the target is a scalar.
|
ANeuralNetworksExecution_getOutputOperandRank
int ANeuralNetworksExecution_getOutputOperandRank( ANeuralNetworksExecution *execution, int32_t index, uint32_t *rank )
Get the dimensional information of the specified output operand of the model of the ANeuralNetworksExecution.
On asynchronous execution initiated by ANeuralNetworksExecution_startCompute, ANeuralNetworksEvent_wait must be called prior to this function to recuperate the resources used by the execution.
Available since API level 29.
Details | |||||||
---|---|---|---|---|---|---|---|
Parameters |
|
||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_OUTPUT_INSUFFICIENT_SIZE if the target output is provided an insufficient buffer at execution time, ANEURALNETWORKS_BAD_DATA if the index is invalid.
|
ANeuralNetworksExecution_setInput
int ANeuralNetworksExecution_setInput( ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const void *buffer, size_t length )
Associate a user buffer with an input of the model of the ANeuralNetworksExecution.
Evaluation of the execution must not have been scheduled. Once evaluation of the execution has been scheduled, the application must not change the content of the buffer until the execution has completed. Evaluation of the execution will not change the content of the buffer.
The provided buffer must outlive the execution.
If the input is optional, you can indicate that it is omitted by passing nullptr for buffer and 0 for length.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the name is not recognized or the buffer is too small for the input.
|
ANeuralNetworksExecution_setInputFromMemory
int ANeuralNetworksExecution_setInputFromMemory( ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const ANeuralNetworksMemory *memory, size_t offset, size_t length )
Associate a region of a memory object with an input of the model of the ANeuralNetworksExecution.
Evaluation of the execution must not have been scheduled. Once evaluation of the execution has been scheduled, the application must not change the content of the region until the execution has completed. Evaluation of the execution will not change the content of the region.
The provided memory must outlive the execution.
If the input is optional, you can indicate that it is omitted by using ANeuralNetworksExecution_setInput instead, passing nullptr for buffer and 0 for length.
See ANeuralNetworksExecution for information on multithreaded usage. See ANeuralNetworksMemory_createFromAHardwareBuffer for information on AHardwareBuffer usage.
Available since API level 27.
Details | |||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the name is not recognized or the buffer is too small for the input.
|
ANeuralNetworksExecution_setMeasureTiming
int ANeuralNetworksExecution_setMeasureTiming( ANeuralNetworksExecution *execution, bool measure )
Specifies whether duration of the ANeuralNetworksExecution is to be measured.
Evaluation of the execution must not have been scheduled.
By default, duration is not measured.
The ANeuralNetworksExecution must have been created with ANeuralNetworksCompilation_createForDevices with numDevices = 1.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksExecution_setOutput
int ANeuralNetworksExecution_setOutput( ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, void *buffer, size_t length )
Associate a user buffer with an output of the model of the ANeuralNetworksExecution.
Evaluation of the execution must not have been scheduled. Once evaluation of the execution has been scheduled, the application must not change the content of the buffer until the execution has completed.
If the output is optional, you can indicate that it is omitted by passing nullptr for buffer and 0 for length.
The provided buffer must outlive the execution.
See ANeuralNetworksExecution for information on multithreaded usage.
Available since API level 27.
Details | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the name is not recognized or the buffer is too small for the output.
|
ANeuralNetworksExecution_setOutputFromMemory
int ANeuralNetworksExecution_setOutputFromMemory( ANeuralNetworksExecution *execution, int32_t index, const ANeuralNetworksOperandType *type, const ANeuralNetworksMemory *memory, size_t offset, size_t length )
Associate a region of a memory object with an output of the model of the ANeuralNetworksExecution.
Evaluation of the execution must not have been scheduled. Once evaluation of the execution has been scheduled, the application must not change the content of the region until the execution has completed.
If the output is optional, you can indicate that it is omitted by using ANeuralNetworksExecution_setOutput instead, passing nullptr for buffer and 0 for length.
The provided memory must outlive the execution.
See ANeuralNetworksExecution for information on multithreaded usage. See ANeuralNetworksMemory_createFromAHardwareBuffer for information on AHardwareBuffer usage.
Available since API level 27.
Details | |||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful, ANEURALNETWORKS_BAD_DATA if the name is not recognized or the buffer is too small for the output.
|
ANeuralNetworksExecution_startCompute
int ANeuralNetworksExecution_startCompute( ANeuralNetworksExecution *execution, ANeuralNetworksEvent **event )
Schedule asynchronous evaluation of the execution.
Schedules asynchronous evaluation of the execution. Once the model has been applied and the outputs are ready to be consumed, the returned event will be signaled. Use ANeuralNetworksEvent_wait to wait for that event.
ANeuralNetworksEvent_wait must be called to recuperate the resources used by the execution.
See ANeuralNetworksExecution for information on multithreaded usage.
See ANeuralNetworksExecution_compute for synchronous execution. Synchronous execution incurs lower overhead than asynchronous execution.
Available since API level 27.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksMemory_createFromAHardwareBuffer
int ANeuralNetworksMemory_createFromAHardwareBuffer( const AHardwareBuffer *ahwb, ANeuralNetworksMemory **memory )
Creates a shared memory object from an AHardwareBuffer handle.
If the shared memory is backed by an AHardwareBuffer of AHARDWAREBUFFER_FORMAT_BLOB format, it can be used the same way as shared memory created from a file handle. See ANeuralNetworksMemory for a description on how to use this shared memory.
If the shared memory is backed by an AHardwareBuffer of a format other than AHARDWAREBUFFER_FORMAT_BLOB, it can only be used for Model inputs and outputs. When calling ANeuralNetworksExecution_setInputFromMemory or ANeuralNetworksExecution_setOutputFromMemory with the shared memory, both offset and length must be set to zero and the entire memory region will be associated with the specified input or output operand. There is no guarantee that an arbitrary AHardwareBuffer_Format and AHardwareBuffer_UsageFlags combination can be used by arbitrary devices. The execution will fail if the selected set of devices cannot consume the buffer.
Calling ANeuralNetworksModel_setOperandValueFromMemory with shared memory backed by an AHardwareBuffer of a format other than AHARDWAREBUFFER_FORMAT_BLOB is disallowed.
The provided AHardwareBuffer must outlive the ANeuralNetworksMemory object.
Available since API level 29.
See also: AHardwareBuffer
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if the request completed normally.
|
ANeuralNetworksMemory_createFromFd
int ANeuralNetworksMemory_createFromFd( size_t size, int protect, int fd, size_t offset, ANeuralNetworksMemory **memory )
Creates a shared memory object from a file descriptor.
The shared memory is backed by a file descriptor via mmap. See ANeuralNetworksMemory for a description on how to use this shared memory.
Available since API level 27.
Details | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if the request completed normally.
|
ANeuralNetworksMemory_free
void ANeuralNetworksMemory_free( ANeuralNetworksMemory *memory )
Delete a memory object.
Destroys the object used by the run time to keep track of the memory. This will free the underlying actual memory if no other code has open handles to this memory.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksModel_addOperand
int ANeuralNetworksModel_addOperand( ANeuralNetworksModel *model, const ANeuralNetworksOperandType *type )
Add an operand to a model.
The order in which the operands are added is important. The first one added to a model will have the index value 0, the second 1, etc. These indexes are used as operand identifiers in ANeuralNetworksModel_addOperation, ANeuralNetworksModel_identifyInputsAndOutputs, ANeuralNetworksModel_setOperandValue, ANeuralNetworksModel_setOperandValueFromMemory, ANeuralNetworksExecution_setInput, ANeuralNetworksExecution_setInputFromMemory, ANeuralNetworksExecution_setOutput, ANeuralNetworksExecution_setOutputFromMemory and ANeuralNetworksExecution_setOperandValue.
Every operand must be referenced in exactly one of the following ways:
- It is identified as a model input with ANeuralNetworksModel_identifyInputsAndOutputs.
- It is identified as a constant with ANeuralNetworksModel_setOperandValue or ANeuralNetworksModel_setOperandValueFromMemory.
- It is identified as an output of exactly one operation with ANeuralNetworksModel_addOperation.An operand that is identified as a model input or as a constant must not also be identified as a model output with ANeuralNetworksModel_identifyInputsAndOutputs.To build a model that can accommodate inputs of various sizes, as you may want to do for a CNN, leave unspecified the dimensions that will vary at run time. If you do so, fully specify dimensions when calling ANeuralNetworksExecution_setInput or ANeuralNetworksExecution_setInputFromMemory.Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.See ANeuralNetworksModel for information on multithreaded usage.Available since API level 27.
Parameters model
The model to be modified.type
The ANeuralNetworksOperandType that describes the shape of the operand. Neither the ANeuralNetworksOperandType nor the dimensions it points to need to outlive the call to ANeuralNetworksModel_addOperand.Returns ANEURALNETWORKS_NO_ERROR if successful.
ANeuralNetworksModel_addOperation
int ANeuralNetworksModel_addOperation( ANeuralNetworksModel *model, ANeuralNetworksOperationType type, uint32_t inputCount, const uint32_t *inputs, uint32_t outputCount, const uint32_t *outputs )
Add an operation to a model.
The operands specified by inputs and outputs must have been previously added by calls to ANeuralNetworksModel_addOperand.
Details | |||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.
See ANeuralNetworksModel for information on multithreaded usage.
Available since API level 27.
Details | |
---|---|
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_create
int ANeuralNetworksModel_create( ANeuralNetworksModel **model )
Create an empty ANeuralNetworksModel.
This only creates the object. Computation is performed once ANeuralNetworksExecution_compute or ANeuralNetworksExecution_startCompute is invoked.
The model should be constructed with calls to ANeuralNetworksModel_addOperation and ANeuralNetworksModel_addOperand
ANeuralNetworksModel_finish should be called once the model has been fully constructed.
ANeuralNetworksModel_free should be called once the model is no longer needed.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_finish
int ANeuralNetworksModel_finish( ANeuralNetworksModel *model )
Indicate that we have finished modifying a model.
Required before calling ANeuralNetworksCompilation_create and ANeuralNetworksCompilation_createForDevices.
An application must ensure that no other thread uses the model at the same time.
This function must only be called once for a given model.
See ANeuralNetworksModel for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_free
void ANeuralNetworksModel_free( ANeuralNetworksModel *model )
Destroy a model.
The model need not have been finished by a call to ANeuralNetworksModel_finish.
See ANeuralNetworksModel for information on multithreaded usage.
Available since API level 27.
Details | |||
---|---|---|---|
Parameters |
|
ANeuralNetworksModel_getSupportedOperationsForDevices
int ANeuralNetworksModel_getSupportedOperationsForDevices( const ANeuralNetworksModel *model, const ANeuralNetworksDevice *const *devices, uint32_t numDevices, bool *supportedOps )
Get the supported operations for a specified set of devices.
If multiple devices are selected, the supported operation list is a union of supported operations of all selected devices.
Available since API level 29.
Details | |||||||||
---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_identifyInputsAndOutputs
int ANeuralNetworksModel_identifyInputsAndOutputs( ANeuralNetworksModel *model, uint32_t inputCount, const uint32_t *inputs, uint32_t outputCount, const uint32_t *outputs )
Specifies which operands will be the model's inputs and outputs.
Every model must have at least one input and one output.
An operand cannot be used for both input and output. Doing so will return an error.
The operands specified by inputs and outputs must have been previously added by calls to ANeuralNetworksModel_addOperand.
Details | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.
See ANeuralNetworksModel for information on multithreaded usage.
Available since API level 27.
ANeuralNetworksModel_relaxComputationFloat32toFloat16
int ANeuralNetworksModel_relaxComputationFloat32toFloat16( ANeuralNetworksModel *model, bool allow )
Specifies whether ANEURALNETWORKS_TENSOR_FLOAT32 is allowed to be calculated with range and/or precision as low as that of the IEEE 754 16-bit floating-point format.
By default, ANEURALNETWORKS_TENSOR_FLOAT32 must be calculated using at least the range and precision of the IEEE 754 32-bit floating-point format.
Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
Available since API level 28.
See ANeuralNetworksModel for information on multithreaded usage.
ANeuralNetworksModel_setOperandSymmPerChannelQuantParams
int ANeuralNetworksModel_setOperandSymmPerChannelQuantParams( ANeuralNetworksModel *model, int32_t index, const ANeuralNetworksSymmPerChannelQuantParams *channelQuant )
Sets an operand's per channel quantization parameters.
Sets parameters required by a tensor of type ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL. This function must be called for every tensor of type ANEURALNETWORKS_TENSOR_QUANT8_SYMM_PER_CHANNEL before calling ANeuralNetworksModel_finish.
Available since API level 29.
Details | |||||||
---|---|---|---|---|---|---|---|
Parameters |
|
||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_setOperandValue
int ANeuralNetworksModel_setOperandValue( ANeuralNetworksModel *model, int32_t index, const void *buffer, size_t length )
Sets an operand to a constant value.
Values of length smaller or equal to ANEURALNETWORKS_MAX_SIZE_OF_IMMEDIATELY_COPIED_VALUES are immediately copied into the model.
For values of length greater than ANEURALNETWORKS_MAX_SIZE_OF_IMMEDIATELY_COPIED_VALUES, a pointer to the buffer is stored within the model. The application must not change the content of this region until all executions using this model have completed. As the data may be copied during processing, modifying the data after this call yields undefined results. The provided buffer must outlive this model.
For large tensors, using ANeuralNetworksModel_setOperandValueFromMemory is likely to be more efficient.
To indicate that an optional operand should be considered missing, pass nullptr for buffer and 0 for length.
Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.
See ANeuralNetworksModel for information on multithreaded usage.
Available since API level 27.
Details | |||||||||
---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworksModel_setOperandValueFromMemory
int ANeuralNetworksModel_setOperandValueFromMemory( ANeuralNetworksModel *model, int32_t index, const ANeuralNetworksMemory *memory, size_t offset, size_t length )
Sets an operand to a value stored in a memory object.
The content of the memory is not copied. A reference to that memory is stored inside the model. The application must not change the content of the memory region until all executions using this model have completed. As the data may be copied during processing, modifying the data after this call yields undefined results.
The provided memory must outlive this model.
To indicate that an optional operand should be considered missing, use ANeuralNetworksModel_setOperandValue instead, passing nullptr for buffer.
Is disallowed to set an operand value with shared memory backed by an AHardwareBuffer of a format other than AHARDWAREBUFFER_FORMAT_BLOB.
Attempting to modify a model once ANeuralNetworksModel_finish has been called will return an error.
See ANeuralNetworksModel for information on multithreaded usage. See ANeuralNetworksMemory_createFromAHardwareBuffer for information on AHardwareBuffer usage.
Available since API level 27.
Details | |||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Parameters |
|
||||||||||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworks_getDevice
int ANeuralNetworks_getDevice( uint32_t devIndex, ANeuralNetworksDevice **device )
Get the representation of the specified device.
Available since API level 29.
Details | |||||
---|---|---|---|---|---|
Parameters |
|
||||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|
ANeuralNetworks_getDeviceCount
int ANeuralNetworks_getDeviceCount( uint32_t *numDevices )
Get the number of available devices.
Available since API level 29.
Details | |||
---|---|---|---|
Parameters |
|
||
Returns |
ANEURALNETWORKS_NO_ERROR if successful.
|