lstmeval.cpp File Reference

#include "commontraining.h"
#include "genericvector.h"
#include "lstmtester.h"
#include "strngs.h"
#include "tprintf.h"

Go to the source code of this file.

Functions
	STRING_PARAM_FLAG (model, "", "Name of model file (training or recognition)")
	STRING_PARAM_FLAG (traineddata, "", "If model is a training checkpoint, then traineddata must " "be the traineddata file that was given to the trainer")
	STRING_PARAM_FLAG (eval_listfile, "", "File listing sample files in lstmf training format.")
	INT_PARAM_FLAG (max_image_MB, 2000, "Max memory to use for images.")
	INT_PARAM_FLAG (verbosity, 1, "Amount of diagnosting information to output (0-2).")
int	main (int argc, char **argv)

Function Documentation

INT_PARAM_FLAG() [1/2]

INT_PARAM_FLAG	(	max_image_MB	,
		2000	,
		"Max memory to use for images."
	)

INT_PARAM_FLAG() [2/2]

INT_PARAM_FLAG	(	verbosity	,
		1	,
		"Amount of diagnosting information to output (0-2)."
	)

main()

int main	(	int	argc,
		char **	argv
	)

This program reads in a text file consisting of feature samples from a training page in the following format:

   FontName UTF8-char-str xmin ymin xmax ymax page-number
    NumberOfFeatureTypes(N)
      FeatureTypeName1 NumberOfFeatures(M)
         Feature1
         ...
         FeatureM
      FeatureTypeName2 NumberOfFeatures(M)
         Feature1
         ...
         FeatureM
      ...
      FeatureTypeNameN NumberOfFeatures(M)
         Feature1
         ...
         FeatureM
   FontName CharName ...

The result of this program is a binary inttemp file used by the OCR engine.

Parameters

argc	number of command line arguments
argv	array of command line arguments

Returns

none

Note

Exceptions: none

History: Fri Aug 18 08:56:17 1989, DSJ, Created.

History: Mon May 18 1998, Christy Russson, Revistion started.

Definition at line 38 of file lstmeval.cpp.

                                {
  ParseArguments(&argc, &argv);
  if (FLAGS_model.empty()) {
    tprintf("Must provide a --model!\n");
    return 1;
  }
  if (FLAGS_eval_listfile.empty()) {
    tprintf("Must provide a --eval_listfile!\n");
    return 1;
  }
  tesseract::TessdataManager mgr;
  if (!mgr.Init(FLAGS_model.c_str())) {
    if (FLAGS_traineddata.empty()) {
      tprintf("Must supply --traineddata to eval a training checkpoint!\n");
      return 1;
    }
    tprintf("%s is not a recognition model, trying training checkpoint...\n",
            FLAGS_model.c_str());
    if (!mgr.Init(FLAGS_traineddata.c_str())) {
      tprintf("Failed to load language model from %s!\n",
              FLAGS_traineddata.c_str());
      return 1;
    }
    GenericVector<char> model_data;
    if (!tesseract::LoadDataFromFile(FLAGS_model.c_str(), &model_data)) {
      tprintf("Failed to load model from: %s\n", FLAGS_model.c_str());
      return 1;
    }
    mgr.OverwriteEntry(tesseract::TESSDATA_LSTM, &model_data[0],
                       model_data.size());
  }
  tesseract::LSTMTester tester(static_cast<inT64>(FLAGS_max_image_MB) *
                               1048576);
  if (!tester.LoadAllEvalData(FLAGS_eval_listfile.c_str())) {
    tprintf("Failed to load eval data from: %s\n", FLAGS_eval_listfile.c_str());
    return 1;
  }
  double errs = 0.0;
  STRING result =
      tester.RunEvalSync(0, &errs, mgr,
                         /*training_stage (irrelevant)*/ 0, FLAGS_verbosity);
  tprintf("%s\n", result.string());
  return 0;
} /* main */

STRING_PARAM_FLAG() [1/3]

STRING_PARAM_FLAG	(	model	,
		""	,
		"Name of model file (training or recognition)"
	)

STRING_PARAM_FLAG() [2/3]

STRING_PARAM_FLAG	(	traineddata	,
		""	,
		"If model is a training	checkpoint,
		then traineddata must " "be the traineddata file that was given to the trainer"
	)

STRING_PARAM_FLAG() [3/3]

STRING_PARAM_FLAG	(	eval_listfile	,
		""	,
		"File listing sample files in lstmf training format."
	)