mastertrainer_test.cc File Reference

#include "include_gunit.h"
#include "commontraining.h"
#include "errorcounter.h"
#include "log.h"
#include "mastertrainer.h"
#include "shapeclassifier.h"
#include "shapetable.h"
#include "trainingsample.h"
#include "unicharset.h"
#include <string>
#include <utility>
#include <vector>

Go to the source code of this file.

Classes
class	MockClassifier
class	MasterTrainerTest

Functions
	TEST_F (MasterTrainerTest, Il1Test)
	TEST_F (MasterTrainerTest, ErrorCounterTest)

Variables
const double	kMin1lDistance = 0.25

Function Documentation

TEST_F() [1/2]

TEST_F	(	MasterTrainerTest	,
		ErrorCounterTest
	)

Definition at line 247 of file mastertrainer_test.cc.

                                            {
#ifdef DISABLED_LEGACY_ENGINE
  // Skip test because LoadTrainingData is missing.
  GTEST_SKIP();
#else
  // Initialize the master_trainer_ from the saved tmp file.
  LoadMasterTrainer();
  // Add the space character to the shape_table_ if not already present to
  // count junk.
  if (shape_table_->FindShape(0, -1) < 0) {
    shape_table_->AddShape(0, 0);
  }
  // Make a mock classifier.
  auto shape_classifier = std::make_unique<MockClassifier>(shape_table_);
  // Get the accuracy report.
  std::string accuracy_report;
  master_trainer_->TestClassifierOnSamples(tesseract::CT_UNICHAR_TOP1_ERR, 0, false,
                                           shape_classifier.get(), &accuracy_report);
  LOG(INFO) << accuracy_report.c_str();
  std::string result_string = accuracy_report.c_str();
  std::vector<std::string> results = split(result_string, '\t');
  EXPECT_EQ(tesseract::CT_SIZE + 1, results.size());
  int result_values[tesseract::CT_SIZE];
  for (int i = 0; i < tesseract::CT_SIZE; ++i) {
    EXPECT_TRUE(safe_strto32(results[i + 1], &result_values[i]));
  }
  // These tests are more-or-less immune to additions to the number of
  // categories or changes in the training data.
  int num_samples = master_trainer_->GetSamples()->num_raw_samples();
  EXPECT_EQ(kNumCorrect, result_values[tesseract::CT_UNICHAR_TOP_OK]);
  EXPECT_EQ(1, result_values[tesseract::CT_FONT_ATTR_ERR]);
  EXPECT_EQ(kNumTopTopErrs, result_values[tesseract::CT_UNICHAR_TOPTOP_ERR]);
  EXPECT_EQ(kNumTop1Errs, result_values[tesseract::CT_UNICHAR_TOP1_ERR]);
  EXPECT_EQ(kNumTop2Errs, result_values[tesseract::CT_UNICHAR_TOP2_ERR]);
  EXPECT_EQ(kNumTopNErrs, result_values[tesseract::CT_UNICHAR_TOPN_ERR]);
  // Each of the TOPTOP errs also counts as a multi-unichar.
  EXPECT_EQ(kNumTopTopErrs - kNumTop1Errs, result_values[tesseract::CT_OK_MULTI_UNICHAR]);
  EXPECT_EQ(num_samples - kNumNonReject, result_values[tesseract::CT_REJECT]);
  EXPECT_EQ(kNumAnswers, result_values[tesseract::CT_NUM_RESULTS]);
#endif
}

TEST_F() [2/2]

TEST_F	(	MasterTrainerTest	,
		Il1Test
	)

Definition at line 234 of file mastertrainer_test.cc.

                                   {
#ifdef DISABLED_LEGACY_ENGINE
  // Skip test because LoadTrainingData is missing.
  GTEST_SKIP();
#else
  // Initialize the master_trainer_ and load the Arial tr file.
  LoadMasterTrainer();
  VerifyIl1();
#endif
}

Variable Documentation

kMin1lDistance

const double kMin1lDistance = 0.25

Definition at line 141 of file mastertrainer_test.cc.