Base abstract class for dataset readers. More...

#include <reader_base.hpp>

Inheritance diagram for se::Reader:

Public Member Functions
	Reader (const ReaderConfig &c)
	Construct a Reader from a ReaderConfig. More...

virtual	~Reader ()

ReaderStatus	nextData (Image< float > &depth_image)
	Read the next depth image. More...

ReaderStatus	nextData (Image< float > &depth_image, Image< uint32_t > &rgba_image)
	Read the next depth and RGBA images. More...

ReaderStatus	nextData (Image< float > &depth_image, Image< uint32_t > &rgba_image, Eigen::Matrix4f &T_WB)
	Read the next depth and RGBA images and ground truth pose. More...

ReaderStatus	getPose (Eigen::Matrix4f &T_WB, const size_t frame)
	Read the ground truth pose at the provided frame number. More...

virtual void	restart ()=0
	Restart reading from the beginning. More...

virtual std::string	name () const =0
	The name of the reader. More...

bool	good () const
	The state of the reader. More...

size_t	frame () const
	The current frame number. More...

size_t	numFrames () const
	The total number of frames in the current dataset. More...

Eigen::Vector2i	depthImageRes () const
	The dimensions of the depth images. More...

Eigen::Vector2i	RGBAImageRes () const
	The dimensions of the RGBA images. More...

bool	isLiveReader () const
	Whether the reader uses a live camera as input. More...

Static Public Member Functions
static ReaderStatus	mergeStatus (ReaderStatus status_1, ReaderStatus status_2)
	Merge se::ReaderStatus values keeping the worst one. More...

Protected Member Functions
ReaderStatus	readPose (Eigen::Matrix4f &T_WB, const size_t frame, const char delimiter=' ')
	Read the ground truth pose at the provided frame number. More...

ReaderStatus	nextPose (Eigen::Matrix4f &T_WB)
	Read the next ground truth pose. More...

Protected Attributes
std::string	sequence_path_

std::string	ground_truth_file_

std::ifstream	ground_truth_fs_

Eigen::Vector2i	depth_image_res_

Eigen::Vector2i	rgba_image_res_

float	fps_

double	spf_

bool	drop_frames_

int	verbose_

bool	is_live_reader_

ReaderStatus	status_

size_t	frame_
	The frame_ is initialized to SIZE_MAX, so that when first incremented it becomes 0. More...

size_t	num_frames_

Detailed Description

Base abstract class for dataset readers.

The overloaded nextData() functions all increment the frame counter, so only one of them should be called within a single pipeline iteration.

Note

Derived classes should adhere to the following rules:

The se::Reader constructor should be called at the very beginning of all derived class constructors.
se::Reader::restart() should be called at the very beginning of all derived class implementations of restart().
The value of se::Reader::is_live_reader_ should be set appropriately at the derived class constructor.
The value of se::Reader::status_ should only be set in derived class constructors and se::Reader::restart() if applicable.

Constructor & Destructor Documentation

◆ Reader()

se::Reader::Reader ( const ReaderConfig & c )

Construct a Reader from a ReaderConfig.

Note: This constructor should be called in all derived class constructors.

Parameters

[in] c The configuration struct to use.

◆ ~Reader()

virtual se::Reader::~Reader ( )

inlinevirtual

Member Function Documentation

◆ nextData() [1/3]

ReaderStatus se::Reader::nextData ( Image< float > & depth_image )

Read the next depth image.

Note: The frame number is incremented when calling this function.

Parameters

[out] depth_image The next depth image.

Returns: An appropriate status code.

◆ nextData() [2/3]

ReaderStatus se::Reader::nextData	(	Image< float > &	depth_image,
		Image< uint32_t > &	rgba_image
	)

Read the next depth and RGBA images.

Note: The frame number is incremented when calling this function.

Parameters

[out]	depth_image	The next depth image.
[out]	rgba_image	The next RGBA image.

Returns: An appropriate status code.

◆ nextData() [3/3]

ReaderStatus se::Reader::nextData	(	Image< float > &	depth_image,
		Image< uint32_t > &	rgba_image,
		Eigen::Matrix4f &	T_WB
	)

Read the next depth and RGBA images and ground truth pose.

Note: The frame number is incremented when calling this function.

Parameters

[out]	depth_image	The next depth image.
[out]	rgba_image	The next RGBA image.
[out]	T_WB	The next ground truth pose.

Returns: An appropriate status code.

◆ getPose()

ReaderStatus se::Reader::getPose	(	Eigen::Matrix4f &	T_WB,
		const size_t	frame
	)

Read the ground truth pose at the provided frame number.

Each line in the ground truth file should correspond to a single depth/RGBA image pair and have a format
... tx ty tz qx qy qz qw,
that is the pose is encoded in the last 7 columns of the line.

Parameters

[in]	frame	The frame number of the requested ground truth pose.
[out]	T_WB	The ground truth pose.

Returns: An appropriate status code.

◆ restart()

virtual void se::Reader::restart ( )

pure virtual

Restart reading from the beginning.

Note: Although this is a virtual function, it does have a default implementation that resets the Reader protected variables. Reader::restart() (this default implementation) should be called at the very beginning of any implementation of restart().

Implemented in se::OpenNIReader, se::InteriorNetReader, se::TUMReader, se::NewerCollegeReader, and se::RAWReader.

◆ name()

virtual std::string se::Reader::name ( ) const

pure virtual

The name of the reader.

Returns: The name of the reader as a string.

Implemented in se::OpenNIReader, se::InteriorNetReader, se::TUMReader, se::NewerCollegeReader, and se::RAWReader.

◆ good()

bool se::Reader::good ( ) const

The state of the reader.

If good() returns false, there is no need to keep reading frames.

Returns: True if the last call to one of the Reader::next*() functions completed successfully.

◆ frame()

size_t se::Reader::frame ( ) const

The current frame number.

Frame numbering starts from 0, so the number of frames read so far is frame() + 1.

Returns: The number of frames read. Returns SIZE_MAX if no frames have been read yet.

◆ numFrames()

size_t se::Reader::numFrames ( ) const

The total number of frames in the current dataset.

Returns: The total number of frames. Returns 0 if the number of frames is unknown (e.g. for camera input).

◆ depthImageRes()

Eigen::Vector2i se::Reader::depthImageRes ( ) const

The dimensions of the depth images.

Returns: A 2D vector containing the width and height of the images.

◆ RGBAImageRes()

Eigen::Vector2i se::Reader::RGBAImageRes ( ) const

The dimensions of the RGBA images.

Returns: A 2D vector containing the width and height of the images.

◆ isLiveReader()

bool se::Reader::isLiveReader ( ) const

Whether the reader uses a live camera as input.

Returns: True if the input is from a live camera, false otherwise.

◆ mergeStatus()

static ReaderStatus se::Reader::mergeStatus	(	ReaderStatus	status_1,
		ReaderStatus	status_2
	)

static

Merge se::ReaderStatus values keeping the worst one.

Parameters

[in]	status_1	The first se::ReaderStatus to merge.
[in]	status_2	The second se::ReaderStatus to merge.

Returns: The worst of the 2 statuses.

◆ readPose()

ReaderStatus se::Reader::readPose	(	Eigen::Matrix4f &	T_WB,
		const size_t	frame,
		const char	delimiter = `' '`
	)

protected

Read the ground truth pose at the provided frame number.

Each line in the ground truth file should correspond to a single depth/RGBA image pair and have a format
... tx ty tz qx qy qz qw,
that is the pose is encoded in the last 7 columns of the line.

Note: Use getPose(...) to request a pose. It keeps track of the ground_truth_frame_ and ifstream state.

Parameters

[out]	T_WB	The ground truth pose.
[in]	frame	The frame number of the requested ground truth pose.
[in]	delimiter	The character delimiting columns in the file. Defaults to space.

Returns: An appropriate status code.

◆ nextPose()

ReaderStatus se::Reader::nextPose ( Eigen::Matrix4f & T_WB )

protected

Read the next ground truth pose.

Each line in the ground truth file should correspond to a single depth/RGBA image pair and have a format
... tx ty tz qx qy qz qw,
that is the pose is encoded in the last 7 columns of the line.

Note: The frame number is NOT incremented inside this function.

Parameters

[out] T_WB The next ground truth pose.

Returns: An appropriate status code.

Member Data Documentation

◆ sequence_path_

std::string se::Reader::sequence_path_

protected

◆ ground_truth_file_

std::string se::Reader::ground_truth_file_

protected

◆ ground_truth_fs_

std::ifstream se::Reader::ground_truth_fs_

protected

◆ depth_image_res_

Eigen::Vector2i se::Reader::depth_image_res_

protected

◆ rgba_image_res_

Eigen::Vector2i se::Reader::rgba_image_res_

protected

◆ fps_

float se::Reader::fps_

protected

◆ spf_

double se::Reader::spf_

protected

◆ drop_frames_

bool se::Reader::drop_frames_

protected

◆ verbose_

int se::Reader::verbose_

protected

◆ is_live_reader_

bool se::Reader::is_live_reader_

protected

◆ status_

ReaderStatus se::Reader::status_

protected

◆ frame_

size_t se::Reader::frame_

protected

The frame_ is initialized to SIZE_MAX, so that when first incremented it becomes 0.

Unsigned integer overflow is defined behaviour in C/C++ so this is safe to do.

◆ num_frames_

size_t se::Reader::num_frames_

protected

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

app/include/reader_base.hpp

Public Member Functions

Static Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ Reader()

◆ ~Reader()

Member Function Documentation

◆ nextData() [1/3]

◆ nextData() [2/3]

◆ nextData() [3/3]

◆ getPose()

◆ restart()

◆ name()

◆ good()

◆ frame()

◆ numFrames()

◆ depthImageRes()

◆ RGBAImageRes()

◆ isLiveReader()

◆ mergeStatus()

◆ readPose()

◆ nextPose()

Member Data Documentation

◆ sequence_path_

◆ ground_truth_file_

◆ ground_truth_fs_

◆ depth_image_res_

◆ rgba_image_res_

◆ fps_

◆ spf_

◆ drop_frames_

◆ verbose_

◆ is_live_reader_

◆ status_

◆ frame_

◆ num_frames_