tesseract::MasterTrainer Class Reference

#include <mastertrainer.h>

Collaboration diagram for tesseract::MasterTrainer:

Public Member Functions
	MasterTrainer (NormalizationMode norm_mode, bool shape_analysis, bool replicate_samples, int debug_level)
	~MasterTrainer ()
bool	Serialize (FILE *fp) const
void	LoadUnicharset (const char *filename)
void	SetFeatureSpace (const IntFeatureSpace &fs)
void	ReadTrainingSamples (const char *page_name, const FEATURE_DEFS_STRUCT &feature_defs, bool verification)
void	AddSample (bool verification, const char *unichar_str, TrainingSample *sample)
void	LoadPageImages (const char *filename)
void	PostLoadCleanup ()
void	PreTrainingSetup ()
void	SetupMasterShapes ()
void	IncludeJunk ()
void	ReplicateAndRandomizeSamplesIfRequired ()
bool	LoadFontInfo (const char *filename)
bool	LoadXHeights (const char *filename)
bool	AddSpacingInfo (const char *filename)
int	GetFontInfoId (const char *font_name)
int	GetBestMatchingFontInfoId (const char *filename)
const STRING &	GetTRFileName (int index) const
void	SetupFlatShapeTable (ShapeTable *shape_table)
CLUSTERER *	SetupForClustering (const ShapeTable &shape_table, const FEATURE_DEFS_STRUCT &feature_defs, int shape_id, int *num_samples)
void	WriteInttempAndPFFMTable (const UNICHARSET &unicharset, const UNICHARSET &shape_set, const ShapeTable &shape_table, CLASS_STRUCT *float_classes, const char *inttemp_file, const char *pffmtable_file)
const UNICHARSET &	unicharset () const
TrainingSampleSet *	GetSamples ()
const ShapeTable &	master_shapes () const
void	DebugCanonical (const char *unichar_str1, const char *unichar_str2)
void	DisplaySamples (const char *unichar_str1, int cloud_font, const char *unichar_str2, int canonical_font)
void	TestClassifierVOld (bool replicate_samples, ShapeClassifier *test_classifier, ShapeClassifier *old_classifier)
void	TestClassifierOnSamples (CountTypes error_mode, int report_level, bool replicate_samples, ShapeClassifier *test_classifier, STRING *report_string)
double	TestClassifier (CountTypes error_mode, int report_level, bool replicate_samples, TrainingSampleSet *samples, ShapeClassifier *test_classifier, STRING *report_string)
float	ShapeDistance (const ShapeTable &shapes, int s1, int s2)

Private Member Functions
void	ReplaceFragmentedSamples ()
void	ClusterShapes (int min_shapes, int max_shape_unichars, float max_dist, ShapeTable *shape_table)

Private Attributes
NormalizationMode	norm_mode_
UNICHARSET	unicharset_
IntFeatureSpace	feature_space_
TrainingSampleSet	samples_
TrainingSampleSet	junk_samples_
TrainingSampleSet	verify_samples_
ShapeTable	master_shapes_
ShapeTable	flat_shapes_
FontInfoTable	fontinfo_table_
GenericVector< int32_t >	xheights_
int	charsetsize_
bool	enable_shape_analysis_
bool	enable_replication_
int *	fragments_
int	prev_unichar_id_
int	debug_level_
IntFeatureMap	feature_map_
GenericVector< Pix * >	page_images_
GenericVector< STRING >	tr_filenames_

Constructor & Destructor Documentation

MasterTrainer()

tesseract::MasterTrainer::MasterTrainer	(	NormalizationMode	norm_mode,
		bool	shape_analysis,
		bool	replicate_samples,
		int	debug_level
	)

~MasterTrainer()

tesseract::MasterTrainer::~MasterTrainer

(

)

Member Function Documentation

AddSample()

void tesseract::MasterTrainer::AddSample	(	bool	verification,
		const char *	unichar_str,
		TrainingSample *	sample
	)

AddSpacingInfo()

bool tesseract::MasterTrainer::AddSpacingInfo

(

const char *

filename

)

ClusterShapes()

void tesseract::MasterTrainer::ClusterShapes ( int  min_shapes, int  max_shape_unichars, float  max_dist, ShapeTable *  shape_table  )

private

DebugCanonical()

void tesseract::MasterTrainer::DebugCanonical	(	const char *	unichar_str1,
		const char *	unichar_str2
	)

DisplaySamples()

void tesseract::MasterTrainer::DisplaySamples	(	const char *	unichar_str1,
		int	cloud_font,
		const char *	unichar_str2,
		int	canonical_font
	)

GetBestMatchingFontInfoId()

int tesseract::MasterTrainer::GetBestMatchingFontInfoId

(

const char *

filename

)

GetFontInfoId()

int tesseract::MasterTrainer::GetFontInfoId

(

const char *

font_name

)

GetSamples()

TrainingSampleSet* tesseract::MasterTrainer::GetSamples ( )

inline

GetTRFileName()

const STRING& tesseract::MasterTrainer::GetTRFileName ( int  index ) const

inline

IncludeJunk()

void tesseract::MasterTrainer::IncludeJunk

(

)

LoadFontInfo()

bool tesseract::MasterTrainer::LoadFontInfo

(

const char *

filename

)

LoadPageImages()

void tesseract::MasterTrainer::LoadPageImages

(

const char *

filename

)

LoadUnicharset()

void tesseract::MasterTrainer::LoadUnicharset

(

const char *

filename

)

LoadXHeights()

bool tesseract::MasterTrainer::LoadXHeights

(

const char *

filename

)

master_shapes()

const ShapeTable& tesseract::MasterTrainer::master_shapes ( ) const

inline

PostLoadCleanup()

void tesseract::MasterTrainer::PostLoadCleanup

(

)

PreTrainingSetup()

void tesseract::MasterTrainer::PreTrainingSetup

(

)

ReadTrainingSamples()

void tesseract::MasterTrainer::ReadTrainingSamples	(	const char *	page_name,
		const FEATURE_DEFS_STRUCT &	feature_defs,
		bool	verification
	)

ReplaceFragmentedSamples()

void tesseract::MasterTrainer::ReplaceFragmentedSamples ( )

private

ReplicateAndRandomizeSamplesIfRequired()

void tesseract::MasterTrainer::ReplicateAndRandomizeSamplesIfRequired

(

)

Serialize()

bool tesseract::MasterTrainer::Serialize

(

FILE *

fp

)

const

SetFeatureSpace()

void tesseract::MasterTrainer::SetFeatureSpace ( const IntFeatureSpace &  fs )

inline

SetupFlatShapeTable()

void tesseract::MasterTrainer::SetupFlatShapeTable

(

ShapeTable *

shape_table

)

SetupForClustering()

CLUSTERER * tesseract::MasterTrainer::SetupForClustering	(	const ShapeTable &	shape_table,
		const FEATURE_DEFS_STRUCT &	feature_defs,
		int	shape_id,
		int *	num_samples
	)

SetupMasterShapes()

void tesseract::MasterTrainer::SetupMasterShapes

(

)

ShapeDistance()

float tesseract::MasterTrainer::ShapeDistance	(	const ShapeTable &	shapes,
		int	s1,
		int	s2
	)

TestClassifier()

double tesseract::MasterTrainer::TestClassifier	(	CountTypes	error_mode,
		int	report_level,
		bool	replicate_samples,
		TrainingSampleSet *	samples,
		ShapeClassifier *	test_classifier,
		STRING *	report_string
	)

TestClassifierOnSamples()

void tesseract::MasterTrainer::TestClassifierOnSamples	(	CountTypes	error_mode,
		int	report_level,
		bool	replicate_samples,
		ShapeClassifier *	test_classifier,
		STRING *	report_string
	)

TestClassifierVOld()

void tesseract::MasterTrainer::TestClassifierVOld	(	bool	replicate_samples,
		ShapeClassifier *	test_classifier,
		ShapeClassifier *	old_classifier
	)

unicharset()

const UNICHARSET& tesseract::MasterTrainer::unicharset ( ) const

inline

WriteInttempAndPFFMTable()

void tesseract::MasterTrainer::WriteInttempAndPFFMTable	(	const UNICHARSET &	unicharset,
		const UNICHARSET &	shape_set,
		const ShapeTable &	shape_table,
		CLASS_STRUCT *	float_classes,
		const char *	inttemp_file,
		const char *	pffmtable_file
	)

Member Data Documentation

charsetsize_

int tesseract::MasterTrainer::charsetsize_

private

debug_level_

int tesseract::MasterTrainer::debug_level_

private

enable_replication_

bool tesseract::MasterTrainer::enable_replication_

private

enable_shape_analysis_

bool tesseract::MasterTrainer::enable_shape_analysis_

private

feature_map_

IntFeatureMap tesseract::MasterTrainer::feature_map_

private

feature_space_

IntFeatureSpace tesseract::MasterTrainer::feature_space_

private

flat_shapes_

ShapeTable tesseract::MasterTrainer::flat_shapes_

private

fontinfo_table_

FontInfoTable tesseract::MasterTrainer::fontinfo_table_

private

fragments_

int* tesseract::MasterTrainer::fragments_

private

junk_samples_

TrainingSampleSet tesseract::MasterTrainer::junk_samples_

private

master_shapes_

ShapeTable tesseract::MasterTrainer::master_shapes_

private

norm_mode_

NormalizationMode tesseract::MasterTrainer::norm_mode_

private

page_images_

GenericVector<Pix*> tesseract::MasterTrainer::page_images_

private

prev_unichar_id_

int tesseract::MasterTrainer::prev_unichar_id_

private

samples_

TrainingSampleSet tesseract::MasterTrainer::samples_

private

tr_filenames_

GenericVector<STRING> tesseract::MasterTrainer::tr_filenames_

private

unicharset_

UNICHARSET tesseract::MasterTrainer::unicharset_

private

verify_samples_

TrainingSampleSet tesseract::MasterTrainer::verify_samples_

private

xheights_

GenericVector<int32_t> tesseract::MasterTrainer::xheights_

private

---

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

• /home/stephane/src/tesseract/src/classify/mastertrainer.h
• /home/stephane/src/tesseract/src/classify/mastertrainer.cpp