ArmNN  NotReleased
BatchNormalizationLayer Class Reference

This layer represents a batch normalization operation. More...

#include <BatchNormalizationLayer.hpp>

Inheritance diagram for BatchNormalizationLayer:
LayerWithParameters< BatchNormalizationDescriptor > Layer IConnectableLayer

Public Member Functions

virtual std::unique_ptr< IWorkloadCreateWorkload (const IWorkloadFactory &factory) const override
 
BatchNormalizationLayerClone (Graph &graph) const override
 
void ValidateTensorShapesFromInputs () override
 
void Accept (ILayerVisitor &visitor) const override
 
- Public Member Functions inherited from LayerWithParameters< BatchNormalizationDescriptor >
const BatchNormalizationDescriptorGetParameters () const
 
void SerializeLayerParameters (ParameterStringifyFunction &fn) const override
 
- Public Member Functions inherited from Layer
 Layer (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, const char *name)
 
 Layer (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, DataLayout layout, const char *name)
 
const std::string & GetNameStr () const
 
const OutputHandlerGetOutputHandler (unsigned int i=0) const
 
OutputHandlerGetOutputHandler (unsigned int i=0)
 
const std::vector< InputSlot > & GetInputSlots () const
 
const std::vector< OutputSlot > & GetOutputSlots () const
 
std::vector< InputSlot >::iterator BeginInputSlots ()
 
std::vector< InputSlot >::iterator EndInputSlots ()
 
std::vector< OutputSlot >::iterator BeginOutputSlots ()
 
std::vector< OutputSlot >::iterator EndOutputSlots ()
 
bool IsOutputUnconnected ()
 
void ResetPriority () const
 
LayerPriority GetPriority () const
 
LayerType GetType () const
 
DataType GetDataType () const
 
const BackendIdGetBackendId () const
 
void SetBackendId (const BackendId &id)
 
virtual void CreateTensorHandles (const TensorHandleFactoryRegistry &registry, const IWorkloadFactory &factory, const bool IsMemoryManaged=true)
 
void VerifyLayerConnections (unsigned int expectedConnections, const CheckLocation &location) const
 
std::vector< TensorShapeInferOutputShapes (const std::vector< TensorShape > &inputShapes) const override
 
virtual void ReleaseConstantData ()
 
template<typename Op >
void OperateOnConstantTensors (Op op)
 
const char * GetName () const override
 
unsigned int GetNumInputSlots () const override
 
unsigned int GetNumOutputSlots () const override
 
const InputSlotGetInputSlot (unsigned int index) const override
 
InputSlotGetInputSlot (unsigned int index) override
 
const OutputSlotGetOutputSlot (unsigned int index=0) const override
 
OutputSlotGetOutputSlot (unsigned int index=0) override
 
void SetGuid (LayerGuid guid)
 
LayerGuid GetGuid () const final
 
void AddRelatedLayerName (const std::string layerName)
 
const std::list< std::string > & GetRelatedLayerNames ()
 
virtual void Reparent (Graph &dest, std::list< Layer *>::const_iterator iterator)=0
 

Public Attributes

std::unique_ptr< ScopedCpuTensorHandlem_Mean
 A unique pointer to store Mean values. More...
 
std::unique_ptr< ScopedCpuTensorHandlem_Variance
 A unique pointer to store Variance values. More...
 
std::unique_ptr< ScopedCpuTensorHandlem_Beta
 A unique pointer to store Beta values. More...
 
std::unique_ptr< ScopedCpuTensorHandlem_Gamma
 A unique pointer to store Gamma values. More...
 

Protected Member Functions

 BatchNormalizationLayer (const BatchNormalizationDescriptor &param, const char *name)
 
 ~BatchNormalizationLayer ()=default
 Default destructor. More...
 
ConstantTensors GetConstantTensorsByRef () override
 
- Protected Member Functions inherited from LayerWithParameters< BatchNormalizationDescriptor >
 LayerWithParameters (unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, const BatchNormalizationDescriptor &param, const char *name)
 
 ~LayerWithParameters ()=default
 
WorkloadInfo PrepInfoAndDesc (QueueDescriptor &descriptor) const
 Helper function to reduce duplication in *LayerCreateWorkload. More...
 
- Protected Member Functions inherited from Layer
virtual ~Layer ()=default
 
template<typename QueueDescriptor >
void CollectQueueDescriptorInputs (QueueDescriptor &descriptor, WorkloadInfo &info) const
 
template<typename QueueDescriptor >
void CollectQueueDescriptorOutputs (QueueDescriptor &descriptor, WorkloadInfo &info) const
 
template<typename QueueDescriptor >
WorkloadInfo PrepInfoAndDesc (QueueDescriptor &descriptor) const
 Helper function to reduce duplication in *LayerCreateWorkload. More...
 
template<typename LayerType , typename ... Params>
LayerTypeCloneBase (Graph &graph, Params &&... params) const
 
- Protected Member Functions inherited from IConnectableLayer
 ~IConnectableLayer ()
 Objects are not deletable via the handle. More...
 

Additional Inherited Members

- Public Types inherited from LayerWithParameters< BatchNormalizationDescriptor >
using DescriptorType = BatchNormalizationDescriptor
 
- Protected Types inherited from Layer
using ConstantTensors = std::vector< std::reference_wrapper< std::unique_ptr< ScopedCpuTensorHandle > >>
 
- Protected Attributes inherited from LayerWithParameters< BatchNormalizationDescriptor >
BatchNormalizationDescriptor m_Param
 The parameters for the layer (not including tensor-valued weights etc.). More...
 
- Protected Attributes inherited from Layer
std::vector< OutputHandlerm_OutputHandlers
 

Detailed Description

This layer represents a batch normalization operation.

Definition at line 15 of file BatchNormalizationLayer.hpp.

Constructor & Destructor Documentation

◆ BatchNormalizationLayer()

BatchNormalizationLayer ( const BatchNormalizationDescriptor param,
const char *  name 
)
protected

Constructor to create a BatchNormalizationLayer.

Parameters
[in]paramBatchNormalizationDescriptor to configure the batch normalization operation.
[in]nameOptional name for the layer.

Definition at line 16 of file BatchNormalizationLayer.cpp.

References armnn::BatchNormalization.

18 {
19 }
LayerWithParameters(unsigned int numInputSlots, unsigned int numOutputSlots, LayerType type, const BatchNormalizationDescriptor &param, const char *name)

◆ ~BatchNormalizationLayer()

~BatchNormalizationLayer ( )
protecteddefault

Default destructor.

Member Function Documentation

◆ Accept()

void Accept ( ILayerVisitor visitor) const
overridevirtual

Implements IConnectableLayer.

Definition at line 71 of file BatchNormalizationLayer.cpp.

References Layer::GetName(), LayerWithParameters< BatchNormalizationDescriptor >::GetParameters(), BatchNormalizationLayer::m_Beta, BatchNormalizationLayer::m_Gamma, BatchNormalizationLayer::m_Mean, BatchNormalizationLayer::m_Variance, and ILayerVisitor::VisitBatchNormalizationLayer().

72 {
73  ConstTensor meanTensor(m_Mean->GetTensorInfo(), m_Mean->Map(true));
74  ConstTensor varianceTensor(m_Variance->GetTensorInfo(), m_Variance->Map(true));
75  ConstTensor betaTensor(m_Beta->GetTensorInfo(), m_Beta->Map(true));
76  ConstTensor gammaTensor(m_Gamma->GetTensorInfo(), m_Gamma->Map(true));
77  visitor.VisitBatchNormalizationLayer(
78  this, GetParameters(), meanTensor, varianceTensor, betaTensor, gammaTensor, GetName());
79 }
const char * GetName() const override
Definition: Layer.hpp:305
std::unique_ptr< ScopedCpuTensorHandle > m_Gamma
A unique pointer to store Gamma values.
const BatchNormalizationDescriptor & GetParameters() const
std::unique_ptr< ScopedCpuTensorHandle > m_Variance
A unique pointer to store Variance values.
std::unique_ptr< ScopedCpuTensorHandle > m_Mean
A unique pointer to store Mean values.
std::unique_ptr< ScopedCpuTensorHandle > m_Beta
A unique pointer to store Beta values.

◆ Clone()

BatchNormalizationLayer * Clone ( Graph graph) const
overridevirtual

Creates a dynamically-allocated copy of this layer.

Parameters
[in]graphThe graph into which this layer is being cloned.

Implements Layer.

Definition at line 39 of file BatchNormalizationLayer.cpp.

References Layer::GetName(), BatchNormalizationLayer::m_Beta, BatchNormalizationLayer::m_Gamma, BatchNormalizationLayer::m_Mean, LayerWithParameters< BatchNormalizationDescriptor >::m_Param, and BatchNormalizationLayer::m_Variance.

40 {
41  auto layer = CloneBase<BatchNormalizationLayer>(graph, m_Param, GetName());
42 
43  layer->m_Mean = m_Mean ? std::make_unique<ScopedCpuTensorHandle>(*m_Mean) : nullptr;
44  layer->m_Variance = m_Variance ? std::make_unique<ScopedCpuTensorHandle>(*m_Variance) : nullptr;
45  layer->m_Beta = m_Beta ? std::make_unique<ScopedCpuTensorHandle>(*m_Beta) : nullptr;
46  layer->m_Gamma = m_Gamma ? std::make_unique<ScopedCpuTensorHandle>(*m_Gamma) : nullptr;
47 
48  return std::move(layer);
49 }
const char * GetName() const override
Definition: Layer.hpp:305
std::unique_ptr< ScopedCpuTensorHandle > m_Gamma
A unique pointer to store Gamma values.
std::unique_ptr< ScopedCpuTensorHandle > m_Variance
A unique pointer to store Variance values.
BatchNormalizationDescriptor m_Param
The parameters for the layer (not including tensor-valued weights etc.).
std::unique_ptr< ScopedCpuTensorHandle > m_Mean
A unique pointer to store Mean values.
std::unique_ptr< ScopedCpuTensorHandle > m_Beta
A unique pointer to store Beta values.

◆ CreateWorkload()

std::unique_ptr< IWorkload > CreateWorkload ( const IWorkloadFactory factory) const
overridevirtual

Makes a workload for the BatchNormalization type.

Parameters
[in]graphThe graph where this layer can be found.
[in]factoryThe workload factory which will create the workload.
Returns
A pointer to the created workload, or nullptr if not created.

Implements Layer.

Definition at line 21 of file BatchNormalizationLayer.cpp.

References IWorkloadFactory::CreateBatchNormalization(), BatchNormalizationLayer::m_Beta, BatchNormalizationQueueDescriptor::m_Beta, BatchNormalizationLayer::m_Gamma, BatchNormalizationQueueDescriptor::m_Gamma, BatchNormalizationLayer::m_Mean, BatchNormalizationQueueDescriptor::m_Mean, BatchNormalizationLayer::m_Variance, BatchNormalizationQueueDescriptor::m_Variance, and LayerWithParameters< BatchNormalizationDescriptor >::PrepInfoAndDesc().

22 {
23  // on this level constant data should not be released..
24  BOOST_ASSERT_MSG(m_Mean != nullptr, "BatchNormalizationLayer: Mean data should not be null.");
25  BOOST_ASSERT_MSG(m_Variance != nullptr, "BatchNormalizationLayer: Variance data should not be null.");
26  BOOST_ASSERT_MSG(m_Beta != nullptr, "BatchNormalizationLayer: Beta data should not be null.");
27  BOOST_ASSERT_MSG(m_Gamma != nullptr, "BatchNormalizationLayer: Gamma data should not be null.");
28 
29  BatchNormalizationQueueDescriptor descriptor;
30 
31  descriptor.m_Mean = m_Mean.get();
32  descriptor.m_Variance = m_Variance.get();
33  descriptor.m_Beta = m_Beta.get();
34  descriptor.m_Gamma = m_Gamma.get();
35 
36  return factory.CreateBatchNormalization(descriptor, PrepInfoAndDesc(descriptor));
37 }
std::unique_ptr< ScopedCpuTensorHandle > m_Gamma
A unique pointer to store Gamma values.
std::unique_ptr< ScopedCpuTensorHandle > m_Variance
A unique pointer to store Variance values.
WorkloadInfo PrepInfoAndDesc(QueueDescriptor &descriptor) const
Helper function to reduce duplication in *LayerCreateWorkload.
std::unique_ptr< ScopedCpuTensorHandle > m_Mean
A unique pointer to store Mean values.
std::unique_ptr< ScopedCpuTensorHandle > m_Beta
A unique pointer to store Beta values.

◆ GetConstantTensorsByRef()

Layer::ConstantTensors GetConstantTensorsByRef ( )
overrideprotectedvirtual

Retrieve the handles to the constant values stored by the layer.

Returns
A vector of the constant tensors stored by this layer.

Reimplemented from Layer.

Definition at line 66 of file BatchNormalizationLayer.cpp.

References BatchNormalizationLayer::m_Beta, BatchNormalizationLayer::m_Gamma, BatchNormalizationLayer::m_Mean, and BatchNormalizationLayer::m_Variance.

67 {
68  return {m_Mean, m_Variance, m_Beta, m_Gamma};
69 }
std::unique_ptr< ScopedCpuTensorHandle > m_Gamma
A unique pointer to store Gamma values.
std::unique_ptr< ScopedCpuTensorHandle > m_Variance
A unique pointer to store Variance values.
std::unique_ptr< ScopedCpuTensorHandle > m_Mean
A unique pointer to store Mean values.
std::unique_ptr< ScopedCpuTensorHandle > m_Beta
A unique pointer to store Beta values.

◆ ValidateTensorShapesFromInputs()

void ValidateTensorShapesFromInputs ( )
overridevirtual

Check if the input tensor shape(s) will lead to a valid configuration of BatchNormalizationLayer.

Implements Layer.

Definition at line 51 of file BatchNormalizationLayer.cpp.

References CHECK_LOCATION, InputSlot::GetConnection(), Layer::GetInputSlot(), Layer::GetOutputSlot(), TensorInfo::GetShape(), IOutputSlot::GetTensorInfo(), OutputSlot::GetTensorInfo(), Layer::InferOutputShapes(), and Layer::VerifyLayerConnections().

52 {
54 
55  auto inferredShapes = InferOutputShapes({ GetInputSlot(0).GetConnection()->GetTensorInfo().GetShape() });
56 
57  BOOST_ASSERT(inferredShapes.size() == 1);
58 
59  ConditionalThrowIfNotEqual<LayerValidationException>(
60  "BatchNormalizationLayer: TensorShape set on OutputSlot[0] does not match the inferred shape.",
62  inferredShapes[0]);
63 
64 }
std::vector< TensorShape > InferOutputShapes(const std::vector< TensorShape > &inputShapes) const override
Definition: Layer.cpp:370
virtual const TensorInfo & GetTensorInfo() const =0
#define CHECK_LOCATION()
Definition: Exceptions.hpp:169
const IOutputSlot * GetConnection() const override
Definition: Layer.hpp:199
void VerifyLayerConnections(unsigned int expectedConnections, const CheckLocation &location) const
Definition: Layer.cpp:337
const TensorShape & GetShape() const
Definition: Tensor.hpp:88
const TensorInfo & GetTensorInfo() const override
Definition: Layer.cpp:63
const OutputSlot & GetOutputSlot(unsigned int index=0) const override
Definition: Layer.hpp:312
const InputSlot & GetInputSlot(unsigned int index) const override
Definition: Layer.hpp:310

Member Data Documentation

◆ m_Beta

◆ m_Gamma

◆ m_Mean

◆ m_Variance


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