Package com.itextpdf.pdf2data.ocr.engine

Class Pdf2DataTATRPostProcessorStaticInitializer

java.lang.Object

com.itextpdf.pdf2data.ocr.engine.Pdf2DataTATRPostProcessorStaticInitializer

       public final class Pdf2DataTATRPostProcessorStaticInitializer extends Object
      

Class for Microsoft TATR models initializing. These models will be used for Pdf2DataTATRPostProcessor.

Method Summary

Modifier and Type	Method	Description
static ai.djl.inference.PredictorList>>	getTableDetectionPredictor()	Creates new Predictor for table detection model.
static ai.djl.inference.PredictorList>>	getTableStructurePredictor()	Creates new Predictor for table structure model.
static void	initializeStaticModels(File tableDetectionModel, File tableStructureModel)	Initialize models for table post-processing.
static boolean	isInitialized()	Checks whether TATR post-preprocessing was enabled or not.
static void	unloadModels()	Close all table models related resources.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

initializeStaticModels

             public static void initializeStaticModels
             (File tableDetectionModel, File tableStructureModel)
            

Initialize models for table post-processing. Note that Pdf2DataTATRPostProcessor will be only created and used afterward if you provide both models file. If any of arguments is null than Pdf2DataTATRPostProcessor will be disabled.

Parameters:

tableDetectionModel - path to your table detection model as File

tableStructureModel - path to your table structure model as File

getTableDetectionPredictor

             public static ai.djl.inference.PredictorList>> getTableDetectionPredictor()
            

Creates new Predictor for table detection model. Note that predictor isn't thread safe, you should create new predictor for each TATR post-processing.

Returns:

Predictor for table detection model

getTableStructurePredictor

             public static ai.djl.inference.PredictorList>> getTableStructurePredictor()
            

Creates new Predictor for table structure model. Note that predictor isn't thread safe, you should create new predictor for each TATR post-processing.

Returns:

Predictor for table structure model

isInitialized

             public static boolean isInitialized()
            

Checks whether TATR post-preprocessing was enabled or not.

Returns:

true if TATR post-preprocessing is enabled, false otherwise

unloadModels

             public static void unloadModels()
            

Close all table models related resources.