Search in sources :

Example 1 with LetterFeature

use of com.joliciel.jochre.letterGuesser.features.LetterFeature in project jochre by urieli.

the class Jochre method doCommandApplyFeatures.

/**
 * Apply a set of features to a given image or a given shape.
 */
public void doCommandApplyFeatures(int imageId, int shapeId, List<String> featureDescriptors) {
    LetterFeatureTester featureTester = new LetterFeatureTester(jochreSession);
    LetterFeatureParser letterFeatureParser = new LetterFeatureParser();
    Set<LetterFeature<?>> features = letterFeatureParser.getLetterFeatureSet(featureDescriptors);
    Set<String> letters = new HashSet<>();
    featureTester.applyFeatures(features, letters, imageId, shapeId);
}
Also used : LetterFeature(com.joliciel.jochre.letterGuesser.features.LetterFeature) LetterFeatureParser(com.joliciel.jochre.letterGuesser.features.LetterFeatureParser) LetterFeatureTester(com.joliciel.jochre.letterGuesser.features.LetterFeatureTester) HashSet(java.util.HashSet)

Example 2 with LetterFeature

use of com.joliciel.jochre.letterGuesser.features.LetterFeature in project jochre by urieli.

the class Jochre method doCommandEvaluateFull.

/**
 * Evaluate a suite of split/merge models and letter guessing model.
 *  @param criteria
 *          for selecting the evaluation corpus
 * @param save
 *          whether or not the letter guesses should be saved
 */
public void doCommandEvaluateFull(CorpusSelectionCriteria criteria, boolean save, File outputDir, MostLikelyWordChooser wordChooser, String suffix, List<DocumentObserver> observers) throws IOException {
    String baseName = jochreSession.getLetterModelPath().substring(0, jochreSession.getLetterModelPath().indexOf("."));
    if (baseName.lastIndexOf("/") > 0)
        baseName = baseName.substring(baseName.lastIndexOf("/") + 1);
    ClassificationModel letterModel = jochreSession.getLetterModel();
    List<String> letterFeatureDescriptors = letterModel.getFeatureDescriptors();
    LetterFeatureParser letterFeatureParser = new LetterFeatureParser();
    Set<LetterFeature<?>> letterFeatures = letterFeatureParser.getLetterFeatureSet(letterFeatureDescriptors);
    LetterGuesser letterGuesser = new LetterGuesser(letterFeatures, letterModel.getDecisionMaker());
    ClassificationModel splitModel = jochreSession.getSplitModel();
    if (splitModel == null)
        throw new IllegalArgumentException("Missing parameter: jochre.image-analyser.split-model");
    List<String> splitFeatureDescriptors = splitModel.getFeatureDescriptors();
    SplitFeatureParser splitFeatureParser = new SplitFeatureParser();
    Set<SplitFeature<?>> splitFeatures = splitFeatureParser.getSplitFeatureSet(splitFeatureDescriptors);
    SplitCandidateFinder splitCandidateFinder = new SplitCandidateFinder(jochreSession);
    splitCandidateFinder.setMinDistanceBetweenSplits(5);
    ShapeSplitter shapeSplitter = new RecursiveShapeSplitter(splitCandidateFinder, splitFeatures, splitModel.getDecisionMaker(), jochreSession);
    ClassificationModel mergeModel = jochreSession.getMergeModel();
    if (mergeModel == null)
        throw new IllegalArgumentException("Missing parameter: jochre.image-analyser.merge-model");
    List<String> mergeFeatureDescriptors = mergeModel.getFeatureDescriptors();
    MergeFeatureParser mergeFeatureParser = new MergeFeatureParser();
    Set<MergeFeature<?>> mergeFeatures = mergeFeatureParser.getMergeFeatureSet(mergeFeatureDescriptors);
    ShapeMerger shapeMerger = new ShapeMerger(mergeFeatures, mergeModel.getDecisionMaker());
    BoundaryDetector boundaryDetector = null;
    String boundaryDetectorTypeName = jochreSession.getConfig().getConfig("jochre.boundaries").getString("boundary-detector-type");
    BoundaryDetectorType boundaryDetectorType = BoundaryDetectorType.valueOf(boundaryDetectorTypeName);
    switch(boundaryDetectorType) {
        case LetterByLetter:
            boundaryDetector = new LetterByLetterBoundaryDetector(shapeSplitter, shapeMerger, jochreSession);
            break;
        case Deterministic:
            boundaryDetector = new DeterministicBoundaryDetector(shapeSplitter, shapeMerger, jochreSession);
            break;
    }
    ImageAnalyser imageAnalyser = new BeamSearchImageAnalyser(boundaryDetector, letterGuesser, wordChooser, jochreSession);
    LetterValidator letterValidator = new ComponentCharacterValidator(jochreSession);
    OriginalShapeLetterAssigner shapeLetterAssigner = new OriginalShapeLetterAssigner();
    shapeLetterAssigner.setEvaluate(true);
    shapeLetterAssigner.setSave(save);
    shapeLetterAssigner.setLetterValidator(letterValidator);
    shapeLetterAssigner.setSingleLetterMethod(false);
    imageAnalyser.addObserver(shapeLetterAssigner);
    ErrorLogger errorLogger = new ErrorLogger(jochreSession);
    Writer errorWriter = null;
    File errorFile = new File(outputDir, baseName + suffix + "errors.txt");
    errorFile.delete();
    errorWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(errorFile, true), "UTF8"));
    errorLogger.setErrorWriter(errorWriter);
    imageAnalyser.addObserver(errorLogger);
    JochreCorpusImageProcessor imageProcessor = new JochreCorpusImageProcessor(criteria, jochreSession);
    imageProcessor.addObserver(imageAnalyser);
    for (DocumentObserver observer : observers) imageProcessor.addObserver(observer);
    imageProcessor.process();
    LOG.debug("F-score for " + jochreSession.getLetterModelPath() + ": " + shapeLetterAssigner.getFScoreCalculator().getTotalFScore());
    String modelFileName = baseName + suffix + "_full";
    File fscoreFile = new File(outputDir, modelFileName + "_fscores.csv");
    Writer fscoreWriter = errorWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(fscoreFile, true), jochreSession.getCsvEncoding()));
    shapeLetterAssigner.getFScoreCalculator().writeScoresToCSV(fscoreWriter);
}
Also used : LetterByLetterBoundaryDetector(com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector) RecursiveShapeSplitter(com.joliciel.jochre.boundaries.RecursiveShapeSplitter) OriginalShapeLetterAssigner(com.joliciel.jochre.analyser.OriginalShapeLetterAssigner) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) ImageAnalyser(com.joliciel.jochre.analyser.ImageAnalyser) DeterministicBoundaryDetector(com.joliciel.jochre.boundaries.DeterministicBoundaryDetector) SplitFeature(com.joliciel.jochre.boundaries.features.SplitFeature) BufferedWriter(java.io.BufferedWriter) MergeFeatureParser(com.joliciel.jochre.boundaries.features.MergeFeatureParser) LetterValidator(com.joliciel.jochre.letterGuesser.LetterValidator) JochreCorpusImageProcessor(com.joliciel.jochre.graphics.JochreCorpusImageProcessor) LetterFeature(com.joliciel.jochre.letterGuesser.features.LetterFeature) LetterFeatureParser(com.joliciel.jochre.letterGuesser.features.LetterFeatureParser) RecursiveShapeSplitter(com.joliciel.jochre.boundaries.RecursiveShapeSplitter) TrainingCorpusShapeSplitter(com.joliciel.jochre.boundaries.TrainingCorpusShapeSplitter) ShapeSplitter(com.joliciel.jochre.boundaries.ShapeSplitter) SplitCandidateFinder(com.joliciel.jochre.boundaries.SplitCandidateFinder) MergeFeature(com.joliciel.jochre.boundaries.features.MergeFeature) OriginalBoundaryDetector(com.joliciel.jochre.boundaries.OriginalBoundaryDetector) BoundaryDetector(com.joliciel.jochre.boundaries.BoundaryDetector) LetterByLetterBoundaryDetector(com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector) DeterministicBoundaryDetector(com.joliciel.jochre.boundaries.DeterministicBoundaryDetector) DocumentObserver(com.joliciel.jochre.doc.DocumentObserver) SplitFeatureParser(com.joliciel.jochre.boundaries.features.SplitFeatureParser) LetterGuesser(com.joliciel.jochre.letterGuesser.LetterGuesser) ErrorLogger(com.joliciel.jochre.analyser.ErrorLogger) TrainingCorpusShapeMerger(com.joliciel.jochre.boundaries.TrainingCorpusShapeMerger) ShapeMerger(com.joliciel.jochre.boundaries.ShapeMerger) FileOutputStream(java.io.FileOutputStream) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) OutputStreamWriter(java.io.OutputStreamWriter) ComponentCharacterValidator(com.joliciel.jochre.letterGuesser.ComponentCharacterValidator) File(java.io.File) ClassificationModel(com.joliciel.talismane.machineLearning.ClassificationModel) UnknownWordListWriter(com.joliciel.jochre.lexicon.UnknownWordListWriter) BufferedWriter(java.io.BufferedWriter) Writer(java.io.Writer) LexiconErrorWriter(com.joliciel.jochre.lexicon.LexiconErrorWriter) OutputStreamWriter(java.io.OutputStreamWriter)

Example 3 with LetterFeature

use of com.joliciel.jochre.letterGuesser.features.LetterFeature in project jochre by urieli.

the class Jochre method doCommandAnalyse.

/**
 * Analyse a set of images based on a given letter-guessing model.
 *
 * @param criteria
 *          the criteria used to select the documents to be analysed
 * @param wordChooser
 *          the word chooser to use
 * @param observers
 *          the observers, used to create analysis output
 */
public void doCommandAnalyse(CorpusSelectionCriteria criteria, MostLikelyWordChooser wordChooser, List<DocumentObserver> observers) throws IOException {
    ClassificationModel letterModel = jochreSession.getLetterModel();
    List<String> letterFeatureDescriptors = letterModel.getFeatureDescriptors();
    LetterFeatureParser letterFeatureParser = new LetterFeatureParser();
    Set<LetterFeature<?>> letterFeatures = letterFeatureParser.getLetterFeatureSet(letterFeatureDescriptors);
    LetterGuesser letterGuesser = new LetterGuesser(letterFeatures, letterModel.getDecisionMaker());
    ImageAnalyser analyser = new BeamSearchImageAnalyser(null, letterGuesser, wordChooser, jochreSession);
    LetterAssigner letterAssigner = new LetterAssigner();
    analyser.addObserver(letterAssigner);
    JochreCorpusImageProcessor imageProcessor = new JochreCorpusImageProcessor(criteria, jochreSession);
    imageProcessor.addObserver(analyser);
    for (DocumentObserver observer : observers) imageProcessor.addObserver(observer);
    imageProcessor.process();
}
Also used : JochreCorpusImageProcessor(com.joliciel.jochre.graphics.JochreCorpusImageProcessor) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) ImageAnalyser(com.joliciel.jochre.analyser.ImageAnalyser) LetterFeature(com.joliciel.jochre.letterGuesser.features.LetterFeature) OriginalShapeLetterAssigner(com.joliciel.jochre.analyser.OriginalShapeLetterAssigner) LetterAssigner(com.joliciel.jochre.analyser.LetterAssigner) DocumentObserver(com.joliciel.jochre.doc.DocumentObserver) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) LetterFeatureParser(com.joliciel.jochre.letterGuesser.features.LetterFeatureParser) ClassificationModel(com.joliciel.talismane.machineLearning.ClassificationModel) LetterGuesser(com.joliciel.jochre.letterGuesser.LetterGuesser)

Example 4 with LetterFeature

use of com.joliciel.jochre.letterGuesser.features.LetterFeature in project jochre by urieli.

the class Jochre method doCommandTrain.

/**
 * Train a letter guessing model.
 *
 * @param featureDescriptors
 *          the feature descriptors for training
 * @param criteria
 *          criteria for selecting images to include when training
 * @param reconstructLetters
 *          whether or not complete letters should be reconstructed for
 *          training, from merged/split letters
 */
public void doCommandTrain(List<String> featureDescriptors, CorpusSelectionCriteria criteria, boolean reconstructLetters) {
    if (jochreSession.getLetterModelPath() == null)
        throw new RuntimeException("Missing argument: letterModel");
    if (featureDescriptors == null)
        throw new JochreException("features is required");
    LetterFeatureParser letterFeatureParser = new LetterFeatureParser();
    Set<LetterFeature<?>> features = letterFeatureParser.getLetterFeatureSet(featureDescriptors);
    BoundaryDetector boundaryDetector = null;
    if (reconstructLetters) {
        ShapeSplitter splitter = new TrainingCorpusShapeSplitter(jochreSession);
        ShapeMerger merger = new TrainingCorpusShapeMerger();
        boundaryDetector = new LetterByLetterBoundaryDetector(splitter, merger, jochreSession);
    } else {
        boundaryDetector = new OriginalBoundaryDetector();
    }
    LetterValidator letterValidator = new ComponentCharacterValidator(jochreSession);
    ClassificationEventStream corpusEventStream = new JochreLetterEventStream(features, boundaryDetector, letterValidator, criteria, jochreSession);
    File letterModelFile = new File(jochreSession.getLetterModelPath());
    letterModelFile.getParentFile().mkdirs();
    ModelTrainerFactory modelTrainerFactory = new ModelTrainerFactory();
    ClassificationModelTrainer trainer = modelTrainerFactory.constructTrainer(jochreSession.getConfig());
    ClassificationModel letterModel = trainer.trainModel(corpusEventStream, featureDescriptors);
    letterModel.persist(letterModelFile);
}
Also used : LetterByLetterBoundaryDetector(com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector) OriginalBoundaryDetector(com.joliciel.jochre.boundaries.OriginalBoundaryDetector) BoundaryDetector(com.joliciel.jochre.boundaries.BoundaryDetector) LetterByLetterBoundaryDetector(com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector) DeterministicBoundaryDetector(com.joliciel.jochre.boundaries.DeterministicBoundaryDetector) TrainingCorpusShapeMerger(com.joliciel.jochre.boundaries.TrainingCorpusShapeMerger) LetterValidator(com.joliciel.jochre.letterGuesser.LetterValidator) ClassificationEventStream(com.joliciel.talismane.machineLearning.ClassificationEventStream) OriginalBoundaryDetector(com.joliciel.jochre.boundaries.OriginalBoundaryDetector) JochreLetterEventStream(com.joliciel.jochre.letterGuesser.JochreLetterEventStream) ModelTrainerFactory(com.joliciel.talismane.machineLearning.ModelTrainerFactory) JochreException(com.joliciel.jochre.utils.JochreException) ClassificationModelTrainer(com.joliciel.talismane.machineLearning.ClassificationModelTrainer) LetterFeature(com.joliciel.jochre.letterGuesser.features.LetterFeature) TrainingCorpusShapeMerger(com.joliciel.jochre.boundaries.TrainingCorpusShapeMerger) ShapeMerger(com.joliciel.jochre.boundaries.ShapeMerger) LetterFeatureParser(com.joliciel.jochre.letterGuesser.features.LetterFeatureParser) TrainingCorpusShapeSplitter(com.joliciel.jochre.boundaries.TrainingCorpusShapeSplitter) RecursiveShapeSplitter(com.joliciel.jochre.boundaries.RecursiveShapeSplitter) TrainingCorpusShapeSplitter(com.joliciel.jochre.boundaries.TrainingCorpusShapeSplitter) ShapeSplitter(com.joliciel.jochre.boundaries.ShapeSplitter) ComponentCharacterValidator(com.joliciel.jochre.letterGuesser.ComponentCharacterValidator) File(java.io.File) ClassificationModel(com.joliciel.talismane.machineLearning.ClassificationModel)

Example 5 with LetterFeature

use of com.joliciel.jochre.letterGuesser.features.LetterFeature in project jochre by urieli.

the class Jochre method doCommandAnalyse.

/**
 * Full analysis, including merge, split and letter guessing.
 *
 * @param pages
 *          the pages to process, empty means all
 */
public void doCommandAnalyse(File sourceFile, MostLikelyWordChooser wordChooser, Set<Integer> pages, List<DocumentObserver> observers, List<PdfImageObserver> imageObservers) throws IOException {
    ClassificationModel letterModel = jochreSession.getLetterModel();
    List<String> letterFeatureDescriptors = letterModel.getFeatureDescriptors();
    LetterFeatureParser letterFeatureParser = new LetterFeatureParser();
    Set<LetterFeature<?>> letterFeatures = letterFeatureParser.getLetterFeatureSet(letterFeatureDescriptors);
    LetterGuesser letterGuesser = new LetterGuesser(letterFeatures, letterModel.getDecisionMaker());
    BoundaryDetector boundaryDetector = null;
    LetterGuessObserver letterGuessObserver = null;
    if (jochreSession.getSplitModel() != null && jochreSession.getMergeModel() != null) {
        boundaryDetector = new DeterministicBoundaryDetector(jochreSession.getSplitModel(), jochreSession.getMergeModel(), jochreSession);
        OriginalShapeLetterAssigner shapeLetterAssigner = new OriginalShapeLetterAssigner();
        shapeLetterAssigner.setEvaluate(false);
        shapeLetterAssigner.setSingleLetterMethod(false);
        letterGuessObserver = shapeLetterAssigner;
    } else {
        boundaryDetector = new OriginalBoundaryDetector();
        LetterAssigner letterAssigner = new LetterAssigner();
        letterGuessObserver = letterAssigner;
    }
    ImageAnalyser analyser = new BeamSearchImageAnalyser(boundaryDetector, letterGuesser, wordChooser, jochreSession);
    analyser.addObserver(letterGuessObserver);
    JochreDocumentGenerator documentGenerator = new JochreDocumentGenerator(sourceFile.getName(), "", jochreSession);
    documentGenerator.addDocumentObserver(analyser);
    for (DocumentObserver observer : observers) documentGenerator.addDocumentObserver(observer);
    if (!sourceFile.exists())
        throw new JochreException("The file " + sourceFile.getPath() + " does not exist");
    if (sourceFile.getName().toLowerCase().endsWith(".pdf")) {
        PdfDocumentProcessor pdfDocumentProcessor = new PdfDocumentProcessor(sourceFile, pages, documentGenerator);
        for (PdfImageObserver imageObserver : imageObservers) {
            pdfDocumentProcessor.addImageObserver(imageObserver);
        }
        pdfDocumentProcessor.process();
    } else if (sourceFile.getName().toLowerCase().endsWith(".png") || sourceFile.getName().toLowerCase().endsWith(".jpg") || sourceFile.getName().toLowerCase().endsWith(".jpeg") || sourceFile.getName().toLowerCase().endsWith(".gif")) {
        ImageDocumentExtractor extractor = new ImageDocumentExtractor(sourceFile, documentGenerator);
        extractor.extractDocument();
    } else if (sourceFile.isDirectory()) {
        ImageDocumentExtractor extractor = new ImageDocumentExtractor(sourceFile, documentGenerator);
        extractor.extractDocument();
    } else {
        throw new RuntimeException("Unrecognised file extension");
    }
}
Also used : PdfImageObserver(com.joliciel.jochre.utils.pdf.PdfImageObserver) OriginalShapeLetterAssigner(com.joliciel.jochre.analyser.OriginalShapeLetterAssigner) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) ImageAnalyser(com.joliciel.jochre.analyser.ImageAnalyser) OriginalBoundaryDetector(com.joliciel.jochre.boundaries.OriginalBoundaryDetector) BoundaryDetector(com.joliciel.jochre.boundaries.BoundaryDetector) LetterByLetterBoundaryDetector(com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector) DeterministicBoundaryDetector(com.joliciel.jochre.boundaries.DeterministicBoundaryDetector) DeterministicBoundaryDetector(com.joliciel.jochre.boundaries.DeterministicBoundaryDetector) OriginalShapeLetterAssigner(com.joliciel.jochre.analyser.OriginalShapeLetterAssigner) LetterAssigner(com.joliciel.jochre.analyser.LetterAssigner) DocumentObserver(com.joliciel.jochre.doc.DocumentObserver) LetterGuesser(com.joliciel.jochre.letterGuesser.LetterGuesser) LetterGuessObserver(com.joliciel.jochre.analyser.LetterGuessObserver) JochreDocumentGenerator(com.joliciel.jochre.doc.JochreDocumentGenerator) PdfDocumentProcessor(com.joliciel.jochre.pdf.PdfDocumentProcessor) OriginalBoundaryDetector(com.joliciel.jochre.boundaries.OriginalBoundaryDetector) JochreException(com.joliciel.jochre.utils.JochreException) LetterFeature(com.joliciel.jochre.letterGuesser.features.LetterFeature) BeamSearchImageAnalyser(com.joliciel.jochre.analyser.BeamSearchImageAnalyser) LetterFeatureParser(com.joliciel.jochre.letterGuesser.features.LetterFeatureParser) ClassificationModel(com.joliciel.talismane.machineLearning.ClassificationModel) ImageDocumentExtractor(com.joliciel.jochre.doc.ImageDocumentExtractor)

Aggregations

LetterFeature (com.joliciel.jochre.letterGuesser.features.LetterFeature)7 LetterFeatureParser (com.joliciel.jochre.letterGuesser.features.LetterFeatureParser)7 ClassificationModel (com.joliciel.talismane.machineLearning.ClassificationModel)6 BeamSearchImageAnalyser (com.joliciel.jochre.analyser.BeamSearchImageAnalyser)5 ImageAnalyser (com.joliciel.jochre.analyser.ImageAnalyser)5 OriginalShapeLetterAssigner (com.joliciel.jochre.analyser.OriginalShapeLetterAssigner)5 BoundaryDetector (com.joliciel.jochre.boundaries.BoundaryDetector)5 DeterministicBoundaryDetector (com.joliciel.jochre.boundaries.DeterministicBoundaryDetector)5 OriginalBoundaryDetector (com.joliciel.jochre.boundaries.OriginalBoundaryDetector)5 LetterGuesser (com.joliciel.jochre.letterGuesser.LetterGuesser)5 LetterAssigner (com.joliciel.jochre.analyser.LetterAssigner)4 LetterByLetterBoundaryDetector (com.joliciel.jochre.boundaries.LetterByLetterBoundaryDetector)4 DocumentObserver (com.joliciel.jochre.doc.DocumentObserver)4 RecursiveShapeSplitter (com.joliciel.jochre.boundaries.RecursiveShapeSplitter)3 ShapeMerger (com.joliciel.jochre.boundaries.ShapeMerger)3 ShapeSplitter (com.joliciel.jochre.boundaries.ShapeSplitter)3 TrainingCorpusShapeMerger (com.joliciel.jochre.boundaries.TrainingCorpusShapeMerger)3 TrainingCorpusShapeSplitter (com.joliciel.jochre.boundaries.TrainingCorpusShapeSplitter)3 JochreCorpusImageProcessor (com.joliciel.jochre.graphics.JochreCorpusImageProcessor)3 ComponentCharacterValidator (com.joliciel.jochre.letterGuesser.ComponentCharacterValidator)3