The MyCaffeConversionControl handles converting between MyCaffe and ONNX formats. The OnnxControl is used to read and write ONNX files. More...

Inheritance diagram for MyCaffe.converter.onnx.MyCaffeConversionControl< T >:

Public Member Functions
	MyCaffeConversionControl ()
	The constructor. More...

	MyCaffeConversionControl (IContainer container)
	The constructor. More...

void	SetWeightScaling (double dfMin, double dfMax)
	Set the scaling factors applied to the weights. More...

void	ConvertMyCaffeToOnnxFile (CudaDnn< T > cuda, Log log, MyCaffeModelData data, string strOutputFile, int nOpSetVersion=9, bool bUseRawData=true, OnnxDefinitions.DataType dstDataType=OnnxDefinitions.DataType.FLOAT)
	Convert a MyCaffe model description, weights and optionally mean image from the MyCaffe model format to the ONNX format and save the result as a .onnx model file. More...

ModelProto	ConvertMyCaffeToOnnx (CudaDnn< T > cuda, Log log, MyCaffeModelData data, int nOpSetVersion=9, bool bUseRawData=true, OnnxDefinitions.DataType dstDataType=OnnxDefinitions.DataType.FLOAT)
	Convert a MyCaffe model description, weights and optionally mean image from the MyCaffe model format to the ONNX format and return the ONNX ModelProto object containing the model in the ONNX format. More...

ModelProto	ConvertMyCaffeToOnnx (MyCaffeControl< T > ctrl, int nOpSetVersion=9, bool bUseRawData=true, OnnxDefinitions.DataType dstDataType=OnnxDefinitions.DataType.FLOAT, Phase phase=Phase.RUN)
	Convert a model currently loaded into the MyCaffeControl to an ONNX ModelProto. More...

void	ConvertMyCaffeToOnnxFile (MyCaffeControl< T > ctrl, string strOnnxFile, int nOpSetVersion=9, bool bUseRawData=true, OnnxDefinitions.DataType dstDataType=OnnxDefinitions.DataType.FLOAT, Phase phase=Phase.RUN)
	Convert a model currently loaded into the MyCaffeControl to an ONNX .onnx model file. More...

MyCaffeModelData	ConvertOnnxToMyCaffeFromFile (CudaDnn< T > cuda, Log log, string strOnnxFile, bool bFixlupNeuronNodes=true, bool bIncludeLastLayerWeights=false, DatasetDescriptor dsTraining=null)
	Convert an ONNX .onnx model file to the MyCaffe model description, weights and optionally mean image. More...

MyCaffeModelData	ConvertOnnxToMyCaffe (CudaDnn< T > cuda, Log log, ModelProto onnxModel, bool bFixupNeuronNodes=true, bool bIncludeLastLayerWeights=false, DatasetDescriptor dsTraining=null)
	Convert an ONNX ModelProto to the MyCaffe model description, weights and optionally mean image. More...

Public Member Functions inherited from System.ComponentModel.Component
	Component ()

Static Public Member Functions
static float[]	getDataAsFloat (TensorProto tensor)
	Converts the tensor data into an array of float. More...

static double[]	getDataAsDouble (TensorProto tensor)
	Converts the tensor data into an array of double. More...

Properties
List< string >	IgnoreLayerNames `[getset]`
	Get/set the list of layer names to ignore (layers are ignored when they contain the text from one of these items). More...

string	ReportString `[get]`
	Returns the report from the conversion. More...

Detailed Description

The MyCaffeConversionControl handles converting between MyCaffe and ONNX formats. The OnnxControl is used to read and write ONNX files.

Template Parameters

T	Specifies the base type used by MyCaffe which is either float or double.

Definition at line 35 of file MyCaffeConversionControl.cs.

Constructor & Destructor Documentation

◆ MyCaffeConversionControl() [1/2]

MyCaffe.converter.onnx.MyCaffeConversionControl< T >.MyCaffeConversionControl ( )

The constructor.

Definition at line 52 of file MyCaffeConversionControl.cs.

◆ MyCaffeConversionControl() [2/2]

MyCaffe.converter.onnx.MyCaffeConversionControl< T >.MyCaffeConversionControl ( IContainer container )

The constructor.

Parameters

container A container holding the component.

Definition at line 62 of file MyCaffeConversionControl.cs.

Member Function Documentation

◆ ConvertMyCaffeToOnnx() [1/2]

ModelProto MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertMyCaffeToOnnx	(	CudaDnn< T >	cuda,
		Log	log,
		MyCaffeModelData	data,
		int	nOpSetVersion = `9`,
		bool	bUseRawData = `true`,
		OnnxDefinitions.DataType	dstDataType = `OnnxDefinitions.DataType.FLOAT`
	)

Convert a MyCaffe model description, weights and optionally mean image from the MyCaffe model format to the ONNX format and return the ONNX ModelProto object containing the model in the ONNX format.

Parameters

cuda	Specifies the connection to cuda uses to interact with the GPU.
log	Specifies the output log used to show progress.
data	Specifies the MyCaffe model data including the model description, the weights and optionally, the image mean.
nOpSetVersion	Specifies the Operation set version (default = 9).
bUseRawData	Optionally, specifies whether or not to store tensor data as RawData or as the native FloatData or DoubleData (default = true).
dstDataType	Optionally, specifies the output data type, which currently can be either FLOAT or DOUBLE (default = FLOAT).

Returns: The model is returned in the ONNX format as a ModelProto (defined within the OnnxControl)

Definition at line 141 of file MyCaffeConversionControl.cs.

◆ ConvertMyCaffeToOnnx() [2/2]

ModelProto MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertMyCaffeToOnnx	(	MyCaffeControl< T >	ctrl,
		int	nOpSetVersion = `9`,
		bool	bUseRawData = `true`,
		OnnxDefinitions.DataType	dstDataType = `OnnxDefinitions.DataType.FLOAT`,
		Phase	phase = `Phase.RUN`
	)

Convert a model currently loaded into the MyCaffeControl to an ONNX ModelProto.

Parameters

ctrl	Specifies the MyCaffeControl object.
nOpSetVersion	Specifies the Operation set version (default = 9).
bUseRawData	Optionally, specifies whether or not to store tensor data as RawData or as the native FloatData or DoubleData (default = true).
dstDataType	Optionally, specifies the output data type, which currently can be either FLOAT or DOUBLE (default = FLOAT).
phase	Optionally, specifies the phase (which netork) to convert (default = RUN).

Returns: The ONNX model proto is returns that matches the network converted.

Definition at line 175 of file MyCaffeConversionControl.cs.

◆ ConvertMyCaffeToOnnxFile() [1/2]

void MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertMyCaffeToOnnxFile	(	CudaDnn< T >	cuda,
		Log	log,
		MyCaffeModelData	data,
		string	strOutputFile,
		int	nOpSetVersion = `9`,
		bool	bUseRawData = `true`,
		OnnxDefinitions.DataType	dstDataType = `OnnxDefinitions.DataType.FLOAT`
	)

Convert a MyCaffe model description, weights and optionally mean image from the MyCaffe model format to the ONNX format and save the result as a .onnx model file.

Parameters

cuda	Specifies the connection to cuda uses to interact with the GPU.
log	Specifies the output log used to show progress.
data	Specifies the MyCaffe model data including the model description, the weights and optionally, the image mean.
strOutputFile	Specifies the .onnx output file.
nOpSetVersion	Specifies the Operation set version (default = 9).
bUseRawData	Optionally, specifies whether or not to store tensor data as RawData or as the native FloatData or DoubleData (default = true).
dstDataType	Optionally, specifies the output data type, which currently can be either FLOAT or DOUBLE (default = FLOAT).

Definition at line 120 of file MyCaffeConversionControl.cs.

◆ ConvertMyCaffeToOnnxFile() [2/2]

void MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertMyCaffeToOnnxFile	(	MyCaffeControl< T >	ctrl,
		string	strOnnxFile,
		int	nOpSetVersion = `9`,
		bool	bUseRawData = `true`,
		OnnxDefinitions.DataType	dstDataType = `OnnxDefinitions.DataType.FLOAT`,
		Phase	phase = `Phase.RUN`
	)

Convert a model currently loaded into the MyCaffeControl to an ONNX .onnx model file.

Parameters

ctrl	Specifies the MyCaffeControl object.
strOnnxFile	Specifies the output .onnx file.
nOpSetVersion	Specifies the Operation set version (default = 9).
bUseRawData	Optionally, specifies whether or not to store tensor data as RawData or as the native FloatData or DoubleData (default = true).
dstDataType	Optionally, specifies the output data type, which currently can be either FLOAT or DOUBLE (default = FLOAT).
phase	Optionally, specifies the phase (which netork) to convert (default = RUN).

Definition at line 190 of file MyCaffeConversionControl.cs.

◆ ConvertOnnxToMyCaffe()

MyCaffeModelData MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertOnnxToMyCaffe	(	CudaDnn< T >	cuda,
		Log	log,
		ModelProto	onnxModel,
		bool	bFixupNeuronNodes = `true`,
		bool	bIncludeLastLayerWeights = `false`,
		DatasetDescriptor	dsTraining = `null`
	)

Convert an ONNX ModelProto to the MyCaffe model description, weights and optionally mean image.

Parameters

cuda	Specifies the connection to cuda uses to interact with the GPU.
log	Specifies the output log used to show progress.
onnxModel	Specifies the ONNX model.
bFixupNeuronNodes	Optionally, specifies to fixup the neuron nodes (e.g. Relu, Prelu, Elu, Sigmoid, Tahn, etc.) by connecting them to inline nodes by connnecting them back to their parent which is common in Caffe type models (default = true).
bIncludeLastLayerWeights	Optionally, specifies to include the weights for the last layer (default = false, usually not included for transfer learning).
dsTraining	Optionally, specifies a training dataset which when supplied converts the model to a training model where inputs are replaced with data layers, and outputs (e.g. softmax) with loss and accuracy layers (default = false).

Returns: The MyCaffe model description, model weights and image mean are returned as a MyCaffeModelData object.

Definition at line 231 of file MyCaffeConversionControl.cs.

◆ ConvertOnnxToMyCaffeFromFile()

MyCaffeModelData MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ConvertOnnxToMyCaffeFromFile	(	CudaDnn< T >	cuda,
		Log	log,
		string	strOnnxFile,
		bool	bFixlupNeuronNodes = `true`,
		bool	bIncludeLastLayerWeights = `false`,
		DatasetDescriptor	dsTraining = `null`
	)

Convert an ONNX .onnx model file to the MyCaffe model description, weights and optionally mean image.

Parameters

cuda	Specifies the connection to cuda uses to interact with the GPU.
log	Specifies the output log used to show progress.
strOnnxFile	Specifies the ONNX .onnx file.
bFixlupNeuronNodes	Optionally, specifies to fixup the neuron nodes (e.g. Relu, Prelu, Elu, Sigmoid, Tahn, etc.) by connecting them to inline nodes by connnecting them back to their parent which is common in Caffe type models (default = true).
bIncludeLastLayerWeights	Optionally, specifies to include the weights for the last layer (default = false, usually not included for transfer learning).
dsTraining	Optionally, specifies a training dataset which when supplied converts the model to a training model where inputs are replaced with data layers, and outputs (e.g. softmax) with loss and accuracy layers (default = false).

Returns: The MyCaffe model description, model weights and image mean are returned as a MyCaffeModelData object.

Definition at line 209 of file MyCaffeConversionControl.cs.

◆ getDataAsDouble()

static double[] MyCaffe.converter.onnx.MyCaffeConversionControl< T >.getDataAsDouble ( TensorProto tensor )

static

Converts the tensor data into an array of double.

Parameters

tensor Specifies the tensor to convert.

Returns: An array of double values is returned.

Definition at line 1793 of file MyCaffeConversionControl.cs.

◆ getDataAsFloat()

static float[] MyCaffe.converter.onnx.MyCaffeConversionControl< T >.getDataAsFloat ( TensorProto tensor )

static

Converts the tensor data into an array of float.

Parameters

tensor Specifies the tensor to convert.

Returns: An array of float values is returned.

Definition at line 1694 of file MyCaffeConversionControl.cs.

◆ SetWeightScaling()

void MyCaffe.converter.onnx.MyCaffeConversionControl< T >.SetWeightScaling	(	double	dfMin,
		double	dfMax
	)

Set the scaling factors applied to the weights.

Parameters

dfMin	Specifies the minimum of the range.
dfMax	Specifies the maximum of the range.

Definition at line 86 of file MyCaffeConversionControl.cs.

Property Documentation

◆ IgnoreLayerNames

List<string> MyCaffe.converter.onnx.MyCaffeConversionControl< T >.IgnoreLayerNames

getset

Get/set the list of layer names to ignore (layers are ignored when they contain the text from one of these items).

Definition at line 95 of file MyCaffeConversionControl.cs.

◆ ReportString

string MyCaffe.converter.onnx.MyCaffeConversionControl< T >.ReportString

get

Returns the report from the conversion.

Definition at line 104 of file MyCaffeConversionControl.cs.

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

C:/Data/Data/SS_Projects/Intelligence/GitHub/MyCaffe/MyCaffe.converter.onnx/MyCaffeConversionControl.cs

Public Member Functions

Static Public Member Functions

Properties

Detailed Description

Constructor & Destructor Documentation

◆ MyCaffeConversionControl() [1/2]

◆ MyCaffeConversionControl() [2/2]

Member Function Documentation

◆ ConvertMyCaffeToOnnx() [1/2]

◆ ConvertMyCaffeToOnnx() [2/2]

◆ ConvertMyCaffeToOnnxFile() [1/2]

◆ ConvertMyCaffeToOnnxFile() [2/2]

◆ ConvertOnnxToMyCaffe()

◆ ConvertOnnxToMyCaffeFromFile()

◆ getDataAsDouble()

◆ getDataAsFloat()

◆ SetWeightScaling()

Property Documentation

◆ IgnoreLayerNames

◆ ReportString