PostProcessing::PostProcessor Class Reference

Process the output of an Inference::InferenceCore More...

#include <post_processor.h>

Public Member Functions
	PostProcessor (float confidence_threshold, IIR::SmoothingSettings smoothing_settings)
	Construct a new Post Processor object. More...
ProcessedResults	run (Inference::InferenceResults inference_core_output)
	Apply post processing to the given Inference::InferenceCore output. More...
bool	set_confidence_threshold (float confidence_threshold)
	Dynamically update the confidence threshold to allow better calibration. More...
float	get_confidence_threshold ()
	Get the confidence threshold object. More...

Detailed Description

Process the output of an Inference::InferenceCore

This class contains some processing steps to make the output of the Inference::InferenceCore more useful. The following operations are performed to process the raw data:

• Disregarding of any predictions below a given confidence threshold
• Smoothing location predictions in time
• Averaging location of left and right body parts if both exceed the confidence threshold

Constructor & Destructor Documentation

PostProcessor()

PostProcessing::PostProcessor::PostProcessor	(	float	confidence_threshold,
		IIR::SmoothingSettings	smoothing_settings
	)

Construct a new Post Processor object.

Parameters

confidence_threshold	Confidence value that must be exceeded for a prediction for a given body part to be useable
smoothing_settings	Settings for the IIR filter

Member Function Documentation

get_confidence_threshold()

float PostProcessing::PostProcessor::get_confidence_threshold

(

)

Get the confidence threshold object.

Returns

float Currently set confidence threshold

run()

ProcessedResults PostProcessing::PostProcessor::run

(

Inference::InferenceResults

inference_core_output

)

Apply post processing to the given Inference::InferenceCore output.

Parameters

inference_core_output

Output from an Inference::InferenceCore being run

Returns

ProcessedResults

set_confidence_threshold()

bool PostProcessing::PostProcessor::set_confidence_threshold

(

float

confidence_threshold

)

Dynamically update the confidence threshold to allow better calibration.

Parameters

confidence_threshold

The updated confidence threshold in the range [0, 1]

Returns

true If updating threshold was successful

false If the new threshold value is invalid

---

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

• post_processor.h
• post_processor.cpp