private void createCnTrainingFile(TrainModel trainModel)
{
logger.Log("Generating cntraining file...");
var mFTrainingExe = new FileInfo(Path.Combine(tesseractDirectory.FullName, "cntraining.exe"));
Process mFTrainingProcess = new Process();
mFTrainingProcess.StartInfo.FileName = mFTrainingExe.FullName;
//tesseractProcess.StartInfo.RedirectStandardOutput = false;
//tesseractProcess.StartInfo.RedirectStandardError = verbose;
mFTrainingProcess.StartInfo.CreateNoWindow = !verbose;
//tesseractProcess.StartInfo.UseShellExecute = false;
//mFTrainingProcess.StartInfo.Arguments = new StringBuilder("--output_trainer ").Append(LOCAL_FOLDER_TEMP).Append(" ").ToString();
foreach (var train in trainModel.Train)
{
StringBuilder args = new StringBuilder(train.FullName);
args.Append(" ");
mFTrainingProcess.StartInfo.Arguments += args.ToString();
}
mFTrainingProcess.Start();
mFTrainingProcess.WaitForExit();
logger.Log("Generation of cntraining file completed");
}
private void createClusteringFile(TrainModel trainModel)
{
logger.Log("Generating clustering file...");
var clusteringExe = new FileInfo(Path.Combine(tesseractDirectory.FullName, "shapeclustering.exe"));
Process shapeClusteringProcess = new Process();
shapeClusteringProcess.StartInfo.FileName = clusteringExe.FullName;
//tesseractProcess.StartInfo.RedirectStandardOutput = false;
//tesseractProcess.StartInfo.RedirectStandardError = verbose;
shapeClusteringProcess.StartInfo.CreateNoWindow = !verbose;
//tesseractProcess.StartInfo.UseShellExecute = false;
//shapeClusteringProcess.StartInfo.Arguments = new StringBuilder("--output_trainer ").Append(Path.Combine(LOCAL_FOLDER_TEMP, "shapetable")).Append(" ").ToString();
shapeClusteringProcess.StartInfo.Arguments += new StringBuilder("-F ").Append(trainModel.FontProperties).Append(" ").ToString();
shapeClusteringProcess.StartInfo.Arguments += new StringBuilder("-U ").Append(trainModel.Unicharset).Append(" ").ToString();
foreach (var train in trainModel.Train)
{
StringBuilder args = new StringBuilder(train.FullName);
args.Append(" ");
shapeClusteringProcess.StartInfo.Arguments += args.ToString();
}
shapeClusteringProcess.Start();
shapeClusteringProcess.WaitForExit();
logger.Log("Generation of clustering file completed");
}
private void createUnicharset(TrainModel trainModel, DirectoryInfo localTempDirectory)
{
logger?.Log("Generating unicharset file...");
var unicharsetExtractorExe = new FileInfo(Path.Combine(tesseractDirectory.FullName, "unicharset_extractor.exe"));
Process unicharsetExtractorProcess = new Process();
unicharsetExtractorProcess.StartInfo.FileName = unicharsetExtractorExe.FullName;
//tesseractProcess.StartInfo.RedirectStandardOutput = false;
//tesseractProcess.StartInfo.RedirectStandardError = verbose;
unicharsetExtractorProcess.StartInfo.CreateNoWindow = !verbose;
//tesseractProcess.StartInfo.UseShellExecute = false;
unicharsetExtractorProcess.StartInfo.Arguments = new StringBuilder("--output_unicharset ").Append(Path.Combine(".", string.Format("unicharset", languageName))).Append(" ").ToString();
foreach (var box in trainModel.Boxes)
{
StringBuilder args = new StringBuilder(box.FullName);
args.Append(" ");
unicharsetExtractorProcess.StartInfo.Arguments += args.ToString();
}
unicharsetExtractorProcess.Start();
unicharsetExtractorProcess.WaitForExit();
trainModel.Unicharset = new FileInfo(localTempDirectory.GetFiles("unicharset")[0].FullName);
logger.Log("Generation of unicharset file completed");
}
private void createTrainFile(TrainModel trainModel, FileInfo tesseractExe)
{
logger?.Log("Generating tr file...");
Process tesseractProcess = new Process();
tesseractProcess.StartInfo.FileName = tesseractExe.FullName;
//tesseractProcess.StartInfo.RedirectStandardOutput = false;
//tesseractProcess.StartInfo.RedirectStandardError = verbose;
tesseractProcess.StartInfo.CreateNoWindow = !verbose;
//tesseractProcess.StartInfo.UseShellExecute = false;
foreach (var image in trainModel.Images)
{
StringBuilder args = new StringBuilder(image.FullName);
args.Append(" ");
args.Append(image.FullName.Substring(0, image.FullName.Length - 4));
args.Append(" box.train");
tesseractProcess.StartInfo.Arguments = args.ToString();
tesseractProcess.Start();
tesseractProcess.WaitForExit();
}
logger?.Log("Generation of tr file completed");
}
private void createFontPropertiesFile(TrainModel trainModel)
{
logger.Log("Generating font properties file...");
logger.Log(string.Format("Insert font properties parameters [{0} 0 0 1 0 0]", fontName), true);
var fontProperties = Console.ReadLine();
if (string.IsNullOrWhiteSpace(fontProperties))
{
fontProperties = string.Format("{0} 0 0 1 0 0", fontName);
}
string fontPropertiesPath = Path.Combine(".", string.Format("{0}.font_properties", languageName));
var stream = new StreamWriter(fontPropertiesPath);
stream.WriteLine(fontProperties);
stream.Close();
stream.Dispose();
trainModel.FontProperties = new FileInfo(fontPropertiesPath);
logger.Log("Generation of font properties file completed");
}
internal void Train()
{
//Generate local folder if not exists
if (!Directory.Exists(LOCAL_FOLDER_TEMP))
{
Directory.CreateDirectory(LOCAL_FOLDER_TEMP);
}
if (!noCopy)
{
Directory.SetCurrentDirectory(LOCAL_FOLDER_TEMP);
}
else
{
Directory.SetCurrentDirectory(imagesDirectory.FullName);
}
// Check the presence of Tesseract exe inside folder
FileInfo tesseractExe = new FileInfo(Path.Combine(tesseractDirectory.FullName, "tesseract.exe"));
if (!tesseractExe.Exists)
{
throw new Exception("Tesseract.exe has not been found inside the provided path. Please provide a valid Tesseract data folder");
}
if (emode == EMode.BOX_CREATE)
{
logger?.Log("Starting process of box creation...", true);
// Retrieve TIF images for training
var images = retrieveTifImages(imagesDirectory);
// Copy TIF images locally
List <FileInfo> localImages = new List <FileInfo>();
if (!noCopy)
{
localImages = copyImagesLocally(images);
}
else
{
localImages = images;
}
// Create boxes
createBoxes(localImages, tesseractExe);
logger?.Log("Box creation completed. Please check they are correct and start training with mode 'train'", true);
}
else if (emode == EMode.TRAIN)
{
logger?.Log("Starting process of training...", true);
var trainModel = new TrainModel();
var localDirectoryInfo = new DirectoryInfo(".");
trainModel.Images = retrieveTifImages(localDirectoryInfo);
trainModel.Boxes = retrieveBoxesFile(localDirectoryInfo);
// Creation of train file
createTrainFile(trainModel, tesseractExe);
trainModel.Train = retrieveTrainFiles(localDirectoryInfo);
// Creation of unicharset file
createUnicharset(trainModel, localDirectoryInfo);
// Creation of font properties file
createFontPropertiesFile(trainModel);
// Creation of clustering file
createClusteringFile(trainModel);
// Creation of mftraining file
createMfTrainingFile(trainModel);
// Creation of cntraining file
createCnTrainingFile(trainModel);
// Creation of unicharambigs file
createUnicharambigs(tesseractExe);
// Rename all file
renameFile(localDirectoryInfo);
// Combine all data in order to generate final training file
combineData();
}
else
{
throw new Exception(string.Format("Invalid mode: {0}", emode.ToString()));
}
}
