Inference::InferenceCore Class Reference

A wrapper to make running inference easier. More...

#include <inference_core.h>

Public Member Functions
	InferenceCore (const char *model_filename, size_t model_input_width, size_t model_input_height)
	Construct a new Inference Core object. More...
InferenceResults	run (PreProcessing::PreProcessedImage preprocessed_image)
	Run a pre-processed image through the loaded model. More...

Detailed Description

A wrapper to make running inference easier.

This class hides some of the complexity of the underlying TensorFlow Lite C++ API. It nonetheless expects preprocessed images. This means images will need to be resized to the model input dimensions and normalised from uint8_t values to floats in the interval [-1..1] before being run through a InferenceCore object.

Constructor & Destructor Documentation

InferenceCore()

Inference::InferenceCore::InferenceCore	(	const char *	model_filename,
		size_t	model_input_width,
		size_t	model_input_height
	)

Construct a new Inference Core object.

Parameters

model_filename	Path to the TensorFlow Lite model
model_input_width	Width of the input to the model
model_input_height	Height of the input to the model

Member Function Documentation

run()

InferenceResults Inference::InferenceCore::run

(

PreProcessing::PreProcessedImage

preprocessed_image

)

Run a pre-processed image through the loaded model.

Parameters

preprocessed_image

An image that has been resized to the model input dimensions and normalised from uint8_t values to floats in the interval [-1..1] before being passed to this function

Returns

InferenceResults

---

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

• inference_core.h
• inference_core.cpp