ONNXTensorView Class Reference

View a Dataset's m_input as ONNX Tensor and also set up output buffers/Tensors. More...

#include <ONNX.h>

Public Member Functions
	ONNXTensorView (Dataset &dataset, int nOutputs)
	Construct a new ONNXTensorView.
Ort::Value *	inputTensor ()
	Get a pointer to the inputTensor.
Ort::Value *	outputTensor ()
	Get a pointer to the outputTensor.
std::vector< float >	outputData ()
	Get a vector of output values (that may have been filled)

Private Attributes
std::vector< int64_t >	m_inputShape
	Shape of the input Tensor.
std::vector< float >	m_outputData
	Output Tensor buffer.
std::vector< int64_t >	m_outputShape
	Shape of the output Tensor.
Ort::MemoryInfo	m_memoryInfo
	MemoryInfo object to be used when constructing the ONNX Tensors - used to specify device (CPU)
Ort::Value	m_inputTensor
	The input Tensor.
Ort::Value	m_outputTensor
	The output Tensor.

Detailed Description

View a Dataset's m_input as ONNX Tensor and also set up output buffers/Tensors.

These views will be passed to Ort::Session::Run

Definition at line 83 of file ONNX.h.

Constructor & Destructor Documentation

ONNXTensorView()

ONNXTensorView ( Dataset & dataset, int nOutputs )

inline

Construct a new ONNXTensorView.

Parameters

dataset	Dataset whose input vector is supposed to be viewed as ONNX tensor
nOutputs	Number of output values

Definition at line 90 of file ONNX.h.

        : m_inputShape{1, dataset.getNumberOfFeatures()}, m_outputData(nOutputs),
          m_outputShape{1, nOutputs}, m_memoryInfo(Ort::MemoryInfo::CreateCpu(
                                                     OrtDeviceAllocator, OrtMemTypeCPU)),
          m_inputTensor(Ort::Value::CreateTensor<float>(
                          m_memoryInfo, dataset.m_input.data(), dataset.m_input.size(),
                          m_inputShape.data(), m_inputShape.size())),
          m_outputTensor(Ort::Value::CreateTensor<float>(
                           m_memoryInfo, m_outputData.data(), m_outputData.size(),
                           m_outputShape.data(), m_outputShape.size())) {}

Member Function Documentation

inputTensor()

Ort::Value * inputTensor ( )

inline

Get a pointer to the inputTensor.

Definition at line 103 of file ONNX.h.

{ return &m_inputTensor; }

outputData()

std::vector< float > outputData ( )

inline

Get a vector of output values (that may have been filled)

Definition at line 113 of file ONNX.h.

{ return m_outputData; }

outputTensor()

Ort::Value * outputTensor ( )

inline

Get a pointer to the outputTensor.

Definition at line 108 of file ONNX.h.

{ return &m_outputTensor; }

Member Data Documentation

m_inputShape

std::vector<int64_t> m_inputShape

private

Shape of the input Tensor.

Definition at line 118 of file ONNX.h.

m_inputTensor

Ort::Value m_inputTensor

private

The input Tensor.

Definition at line 138 of file ONNX.h.

m_memoryInfo

Ort::MemoryInfo m_memoryInfo

private

MemoryInfo object to be used when constructing the ONNX Tensors - used to specify device (CPU)

Definition at line 133 of file ONNX.h.

m_outputData

std::vector<float> m_outputData

private

Output Tensor buffer.

Definition at line 123 of file ONNX.h.

m_outputShape

std::vector<int64_t> m_outputShape

private

Shape of the output Tensor.

Definition at line 128 of file ONNX.h.

m_outputTensor

Ort::Value m_outputTensor

private

The output Tensor.

Definition at line 143 of file ONNX.h.

---

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

• mva/methods/include/ONNX.h