public static void TestClassification(string trainingFolder, string pdfPath)
{
var svm = Serializer.Load <MulticlassSupportVectorMachine <Gaussian> >(Path.Combine(trainingFolder, "model.gz"), SerializerCompression.GZip);
using (var document = PdfDocument.Open(pdfPath))
{
for (var i = 0; i < document.NumberOfPages; i++)
{
var page = document.GetPage(i + 1);
var words = page.GetWords();
if (words.Count() == 0)
{
continue;
}
var blocks = DocstrumBoundingBoxes.Instance.GetBlocks(words);
foreach (var block in blocks)
{
var letters = block.TextLines.SelectMany(li => li.Words).SelectMany(w => w.Letters);
var paths = FeatureHelper.GetPathsInside(block.BoundingBox, page.ExperimentalAccess.Paths);
var images = FeatureHelper.GetImagesInside(block.BoundingBox, page.GetImages());
var features = FeatureHelper.GetFeatures(page, block.BoundingBox, letters, paths, images);
var category = svm.Decide(features);
Console.WriteLine(FeatureHelper.Categories[category]);
Console.WriteLine(block.Text);
Console.WriteLine();
}
Console.ReadKey();
}
}
}
/// <summary>
/// Generate a csv file of features. You will need the pdf documents and the ground truths in PAGE xml format.
/// </summary>
/// <param name="trainingFolder">The path to the training folder. Should contain both the pdf files and their corresponding ground truth xml files.</param>
/// <param name="numberOfPdfDocs">Number of documents to concider.</param>
public static void GenerateCsv(string trainingFolder, int numberOfPdfDocs)
{
List <double[]> features = new List <double[]>();
List <int> categories = new List <int>();
int done = 0;
DirectoryInfo d = new DirectoryInfo(trainingFolder);
var pdfFileLinks = d.GetFiles("*.pdf");
var indexesSelected = GenerateRandom(numberOfPdfDocs, 0, pdfFileLinks.Length);
foreach (int index in indexesSelected)
{
var pdfFile = pdfFileLinks[index];
string fileName = pdfFile.Name;
string xmlFileNameTemplate = fileName.Replace(".pdf", "_*.xml");
var pageXmlLinks = d.GetFiles(xmlFileNameTemplate);
if (pageXmlLinks.Length == 0)
{
Console.BackgroundColor = ConsoleColor.Red;
Console.WriteLine("No PageXml file found for document '" + fileName + "'");
Console.ResetColor();
continue;
}
try
{
using (var doc = PdfDocument.Open(pdfFile.FullName))
{
foreach (var pageXmlLink in pageXmlLinks)
{
var pageXml = Deserialize(pageXmlLink.FullName);
int pageNo = ParseXmlFileName(pageXmlLink.Name);
var page = doc.GetPage(pageNo + 1);
var blocks = pageXml.Page.Items;
foreach (var block in blocks)
{
int category = -1;
PdfRectangle bbox = new PdfRectangle();
if (block is PageXmlTextRegion textBlock)
{
bbox = ParsePageXmlCoord(textBlock.Coords.Points, (double)page.Height);
switch (textBlock.Type)
{
case PageXmlTextSimpleType.Heading:
category = 0;
break;
case PageXmlTextSimpleType.Paragraph:
category = 1;
break;
case PageXmlTextSimpleType.LisLabel:
category = 2;
break;
default:
throw new ArgumentException("Unknown category");
}
}
else if (block is PageXmlTableRegion tableBlock)
{
bbox = ParsePageXmlCoord(tableBlock.Coords.Points, (double)page.Height);
category = 3;
}
else if (block is PageXmlImageRegion imageBlock)
{
bbox = ParsePageXmlCoord(imageBlock.Coords.Points, (double)page.Height);
category = 4;
}
else
{
throw new ArgumentException("Unknown region type");
}
var letters = FeatureHelper.GetLettersInside(bbox, page.Letters).ToList();
var paths = FeatureHelper.GetPathsInside(bbox, page.ExperimentalAccess.Paths).ToList();
var images = FeatureHelper.GetImagesInside(bbox, page.GetImages());
var f = FeatureHelper.GetFeatures(page, bbox, letters, paths, images);
if (category == -1)
{
throw new ArgumentException("Unknown category number.");
}
if (f != null)
{
features.Add(f);
categories.Add(category);
}
}
}
}
}
catch (Exception ex)
{
Console.ForegroundColor = ConsoleColor.Red;
Console.WriteLine("Error for document '" + fileName + "': " + ex.Message);
Console.ResetColor();
}
Console.WriteLine(done++);
}
if (features.Count != categories.Count)
{
throw new ArgumentException("features and categories don't have the same size");
}
string[] csv = features.Zip(categories, (f, c) => string.Join(',', f) + "," + c).ToArray();
File.WriteAllLines(Path.Combine(trainingFolder, "features.csv"), csv);
}