ActsPlugins::MLTrackClassifier Class Reference

Specialized class that labels tracks as good/duplicate/fake using a deep neural network. More...

#include </home/runner/work/acts/acts/Plugins/Onnx/include/ActsPlugins/Onnx/MLTrackClassifier.hpp>

Inheritance diagram for ActsPlugins::MLTrackClassifier:

Collaboration diagram for ActsPlugins::MLTrackClassifier:

Public Types
enum class	TrackLabels { eGood , eDuplicate , eFake }
	The labels for track quality.

Public Member Functions
bool	isDuplicate (std::vector< float > &inputFeatures, double decisionThreshProb) const
	Check if the predicted track label is 'duplicate'.
TrackLabels	predictTrackLabel (std::vector< float > &inputFeatures, double decisionThreshProb) const
	Predict the track label.
Public Member Functions inherited from ActsPlugins::OnnxRuntimeBase
	OnnxRuntimeBase ()=default
	Default constructor.
	OnnxRuntimeBase (Ort::Env &env, const char *modelPath)
	Parametrized constructor.
	~OnnxRuntimeBase ()=default
	Default destructor.
std::vector< std::vector< float > >	runONNXInference (NetworkBatchInput &inputTensorValues) const
	Run the ONNX inference function for a batch of input.
std::vector< float >	runONNXInference (std::vector< float > &inputTensorValues) const
	Run the ONNX inference function.
std::vector< std::vector< std::vector< float > > >	runONNXInferenceMultiOutput (NetworkBatchInput &inputTensorValues) const
	Run the multi-output ONNX inference function for a batch of input.

Detailed Description

Specialized class that labels tracks as good/duplicate/fake using a deep neural network.

Member Function Documentation

isDuplicate()

bool ActsPlugins::MLTrackClassifier::isDuplicate	(	std::vector< float > &	inputFeatures,
		double	decisionThreshProb ) const

Check if the predicted track label is 'duplicate'.

Parameters

inputFeatures	The vector of input features for the trajectory to be classified
decisionThreshProb	The probability threshold used to predict the track label

Returns

If the predicted track label is 'duplicate'

predictTrackLabel()

TrackLabels ActsPlugins::MLTrackClassifier::predictTrackLabel	(	std::vector< float > &	inputFeatures,
		double	decisionThreshProb ) const

Predict the track label.

Parameters

inputFeatures	The vector of input features for the trajectory to be classified
decisionThreshProb	The probability threshold used to predict the track label

Returns

The predicted track label of the trajectory