Create an ImageDescriber interface for Roma deep dense feature extractor. More...

#include <ImageDescriber_Roma.hpp>

Inheritance diagram for aliceVision::feature::ImageDescriber_Roma:

Public Member Functions
bool	useCuda () const override
	Check if the image describer uses CUDA. More...

bool	useFloatImage () const override
	Check if the image describer uses a float image. More...

EImageDescriberType	getDescriberType () const override
	Get the corresponding EImageDescriberType. More...

std::size_t	getMemoryConsumption (std::size_t width, std::size_t height) const override
	Get the total amount of RAM needed for a feature extraction of an image of the given dimension. More...

void	setConfigurationPreset (ConfigurationPreset preset) override
	Use a preset to control the number of detected regions. More...

bool	describe (const image::Image< unsigned char > &image, std::unique_ptr< Regions > &regions, const image::Image< unsigned char > *mask=nullptr) override
	Detect regions on the 8-bit image and compute their attributes. More...

void	allocate (std::unique_ptr< Regions > &regions) const override
	Allocate Regions type for Roma. More...

Public Member Functions inherited from aliceVision::feature::ImageDescriber
virtual void	setUpRight ([[maybe_unused]] bool upRight)
	Set whether the image describer is always upright. More...

virtual void	setUseCuda ([[maybe_unused]] bool useCuda)
	Set whether the image describer needs to use CUDA implementation. More...

virtual void	setCudaPipe ([[maybe_unused]] int pipe)
	Set the CUDA pipe. More...

virtual bool	describe ([[maybe_unused]] const image::Image< unsigned char > &image, [[maybe_unused]] std::unique_ptr< Regions > &regions, [[maybe_unused]] const image::Image< unsigned char > *mask=nullptr)
	Detect regions on the 8-bit image and compute their attributes (description) More...

virtual bool	describe ([[maybe_unused]] const image::Image< float > &image, [[maybe_unused]] std::unique_ptr< Regions > &regions, [[maybe_unused]] const image::Image< unsigned char > *mask=nullptr)
	Detect regions on the float image and compute their attributes (description) More...

void	Load (Regions *regions, const std::string &sfileNameFeats, const std::string &sfileNameDescs) const

void	Save (const Regions *regions, const std::string &sfileNameFeats, const std::string &sfileNameDescs) const

void	LoadFeatures (Regions *regions, const std::string &sfileNameFeats) const

Detailed Description

Create an ImageDescriber interface for Roma deep dense feature extractor.

This is a "virtual" describer: it does not perform extraction itself but provides the AliceVision interface to load/save pre-computed Roma regions. Actual feature extraction is expected to be performed externally (e.g. via a dedicated Python pipeline) and the resulting .feat/.desc files consumed through this describer.

Member Function Documentation

◆ allocate()

void aliceVision::feature::ImageDescriber_Roma::allocate ( std::unique_ptr< Regions > & regions ) const

overridevirtual

Allocate Regions type for Roma.

Parameters

[in,out] regions

Implements aliceVision::feature::ImageDescriber.

◆ describe()

bool aliceVision::feature::ImageDescriber_Roma::describe	(	const image::Image< unsigned char > &	image,
		std::unique_ptr< Regions > &	regions,
		const image::Image< unsigned char > *	mask = `nullptr`
	)

override

Detect regions on the 8-bit image and compute their attributes.

For this virtual describer the function is a no-op stub: Roma features are expected to be pre-computed externally. Calling this will throw.

Parameters

[in]	image	Input 8-bit grayscale image (unused)
[out]	regions	Output regions (unused)
[in]	mask	Optional keypoint mask (unused)

Returns: False – not implemented

◆ getDescriberType()

EImageDescriberType aliceVision::feature::ImageDescriber_Roma::getDescriberType ( ) const

inlineoverridevirtual

Get the corresponding EImageDescriberType.

Returns: EImageDescriberType::ROMA

Implements aliceVision::feature::ImageDescriber.

◆ getMemoryConsumption()

std::size_t aliceVision::feature::ImageDescriber_Roma::getMemoryConsumption	(	std::size_t	width,
		std::size_t	height
	)		const

inlineoverridevirtual

Get the total amount of RAM needed for a feature extraction of an image of the given dimension.

Parameters

[in]	width	The image width
[in]	height	The image height

Returns: total amount of memory needed (conservative upper bound)

Implements aliceVision::feature::ImageDescriber.

◆ setConfigurationPreset()

void aliceVision::feature::ImageDescriber_Roma::setConfigurationPreset ( ConfigurationPreset preset )

overridevirtual

Use a preset to control the number of detected regions.

Parameters

[in] preset The preset configuration

Implements aliceVision::feature::ImageDescriber.

◆ useCuda()

bool aliceVision::feature::ImageDescriber_Roma::useCuda ( ) const

inlineoverridevirtual

Check if the image describer uses CUDA.

Returns: False – extraction is performed externally

Implements aliceVision::feature::ImageDescriber.

◆ useFloatImage()

bool aliceVision::feature::ImageDescriber_Roma::useFloatImage ( ) const

inlineoverridevirtual

Check if the image describer uses a float image.

Returns: False – extraction is performed externally

Implements aliceVision::feature::ImageDescriber.

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

/home/docs/checkouts/readthedocs.org/user_builds/alicevision/checkouts/latest/src/aliceVision/feature/virtual/ImageDescriber_Roma.hpp
/home/docs/checkouts/readthedocs.org/user_builds/alicevision/checkouts/latest/src/aliceVision/feature/virtual/ImageDescriber_Roma.cpp

Public Member Functions

Detailed Description

Member Function Documentation

◆ allocate()

◆ describe()

◆ getDescriberType()

◆ getMemoryConsumption()

◆ setConfigurationPreset()

◆ useCuda()

◆ useFloatImage()