arm_compute::NEInstanceNormalizationLayerEx Class Reference

#include <NEInstanceNormalizationLayerEx.h>

Collaboration diagram for arm_compute::NEInstanceNormalizationLayerEx:

Public Member Functions
	NEInstanceNormalizationLayerEx (std::shared_ptr< IMemoryManager > memory_manager=nullptr)
void	configure (ITensor *input, ITensor *output, ITensor *gamma, ITensor *beta, float epsilon=1e-12f)
void	run () override

Static Public Member Functions
static Status	validate (const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *gamma=nullptr, const ITensorInfo *beta=nullptr, float epsilon=1e-12f)

Detailed Description

Basic function to perform a Instance normalization.

This function runs the following kernels:

1. NEInstanceNormalizationLayerKernelEx

Definition at line 64 of file NEInstanceNormalizationLayerEx.h.

Constructor & Destructor Documentation

NEInstanceNormalizationLayerEx()

arm_compute::NEInstanceNormalizationLayerEx::NEInstanceNormalizationLayerEx

(

std::shared_ptr< IMemoryManager >

memory_manager = nullptr

)

Constructor

Definition at line 48 of file NEInstanceNormalizationLayerEx.cpp.

  : _memory_group(std::move(memory_manager)), _normalization_kernel(), _is_nchw(false),
    _permute_input(), _permute_output(), _permuted_input(), _permuted_output()
{
}

Member Function Documentation

configure()

void arm_compute::NEInstanceNormalizationLayerEx::configure	(	ITensor *	input,
		ITensor *	output,
		ITensor *	gamma,
		ITensor *	beta,
		float	epsilon = 1e-12f
	)

Set the input and output tensors.

Parameters

[in,out]	input	Source tensor. In case of output tensor = nullptr this tensor will store the result of the normalization. Data types supported: F16/F32. Data layout supported: NHWC, NCHW
[out]	output	Destination tensor. Data types and data layouts supported: same as input.
[in]	gamma	(Optional) The scale scalar value applied to the normalized tensor. Defaults to 1.0
[in]	beta	(Optional) The offset scalar value applied to the normalized tensor. Defaults to 0.0
[in]	epsilon	(Optional) Lower bound value for the normalization. Defaults to 1e-12

Definition at line 55 of file NEInstanceNormalizationLayerEx.cpp.

{
  const DataLayout data_layout = input->info()->data_layout();
 
  // Configure Kernels
  _is_nchw = data_layout == DataLayout::NCHW;
 
  if (!_is_nchw)
  {
    _memory_group.manage(&_permuted_input);
    _memory_group.manage(&_permuted_output);
 
    // Configure the function to transform the input tensor from NHWC -> NCHW
    _permute_input.configure(input, &_permuted_input, PermutationVector(1U, 2U, 0U));
    _permuted_input.info()->set_data_layout(DataLayout::NCHW);
 
    _normalization_kernel.configure(&_permuted_input, &_permuted_output, gamma, beta, epsilon);
    _permuted_output.info()->set_data_layout(DataLayout::NCHW);
 
    _permute_output.configure(&_permuted_output, output != nullptr ? output : input,
                              PermutationVector(2U, 0U, 1U));
    _permuted_input.allocator()->allocate();
    _permuted_output.allocator()->allocate();
  }
  else
  {
    _normalization_kernel.configure(input, output, gamma, beta, epsilon);
  }
}

References arm_compute::NEInstanceNormalizationLayerKernelEx::configure().

run()

void arm_compute::NEInstanceNormalizationLayerEx::run ( )

override

Definition at line 95 of file NEInstanceNormalizationLayerEx.cpp.

{
  MemoryGroupResourceScope scope_mg(_memory_group);
 
  // Permute input
  if (!_is_nchw)
  {
    _permute_input.run();
  }
 
  NEScheduler::get().schedule(&_normalization_kernel, Window::DimZ);
 
  // Permute output
  if (!_is_nchw)
  {
    _permute_output.run();
  }
}

Referenced by package.infer.session::inference().

validate()

Status arm_compute::NEInstanceNormalizationLayerEx::validate ( const ITensorInfo *  input, const ITensorInfo *  output, const ITensorInfo *  gamma = nullptr, const ITensorInfo *  beta = nullptr, float  epsilon = 1e-12f  )

static

Static function to check if given info will lead to a valid configuration of NEInstanceNormalizationLayer.

Parameters

[in]	input	Source tensor info. Data types supported: F16/F32. Data layout supported: NHWC, NCHW
[in]	output	Destination tensor info. Data types and data layouts supported: same as input.
[in]	gamma	(Optional) The scale scalar value applied to the normalized tensor. Defaults to 1.0
[in]	beta	(Optional) The offset scalar value applied to the normalized tensor. Defaults to 0.0
[in]	epsilon	(Optional) Lower bound value for the normalization. Defaults to 1e-12

Returns

a status

Definition at line 86 of file NEInstanceNormalizationLayerEx.cpp.

{
  return NEInstanceNormalizationLayerKernelEx::validate(
    &input->clone()->set_data_layout(DataLayout::NCHW),
    &output->clone()->set_data_layout(DataLayout::NCHW), gamma, beta, epsilon);
}

References arm_compute::NEInstanceNormalizationLayerKernelEx::validate().

---

The documentation for this class was generated from the following files:

• compute/ARMComputeEx/arm_compute/runtime/NEON/functions/NEInstanceNormalizationLayerEx.h
• compute/ARMComputeEx/src/runtime/NEON/functions/NEInstanceNormalizationLayerEx.cpp