You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
Go to file
Konstantin Baierer ca23b32e9b split do_order_of_regions, lots of logging 4 years ago
.circleci set up circle ci 4 years ago
.github/workflows ci: install test dependencies 4 years ago
sbb_newspapers_org_image split do_order_of_regions, lots of logging 4 years ago
tests do an actual test run 4 years ago
.gitignore split utils into contour, rotate, separate_lines 4 years ago
LICENSE extend setup.py, add Makefile, gitignore, requirements.txt 4 years ago
Makefile do an actual test run 4 years ago
README.md Update README.md (#9) 4 years ago
requirements-test.txt 🎨 reformat cli.py with black 4 years ago
requirements.txt restrict keras version to < 2.4 4 years ago
setup.py rfct: move CLI to dedicated module 4 years ago

README.md

Eynollah

Document Layout Analysis

Introduction

This tool performs document layout analysis (segmentation) from image data and returns the results as PAGE-XML.

It can currently detect the following layout classes/elements:

In addition, the tool can be used to detect the ReadingOrder of regions. The final goal is to feed the output to an OCR model.

The tool uses a combination of various models and heuristics (see flowchart below for the different stages and how they interact):

The first three stages are based on pixelwise segmentation.

Border detection

For the purpose of text recognition (OCR) and in order to avoid noise being introduced from texts outside the printspace, one first needs to detect the border of the printed frame. This is done by a binary pixelwise-segmentation model trained on a dataset of 2,000 documents where about 1,200 of them come from the dhSegment project (you can download the dataset from here) and the remainder having been annotated in SBB. For border detection, the model needs to be fed with the whole image at once rather than separated in patches.

Layout detection

As a next step, text regions need to be identified by means of layout detection. Again a pixelwise segmentation model was trained on 131 labeled images from the SBB digital collections, including some data augmentation. Since the target of this tool are historical documents, we consider as main region types text regions, separators, images, tables and background - each with their own subclasses, e.g. in the case of text regions, subclasses like header/heading, drop capital, main body text etc. While it would be desirable to detect and classify each of these classes in a granular way, there are also limitations due to having a suitably large and balanced training set. Accordingly, the current version of this tool is focussed on the main region types background, text region, image and separator.

Textline detection

In a subsequent step, binary pixelwise segmentation is used again to classify pixels in a document that constitute textlines. For textline segmentation, a model was initially trained on documents with only one column/block of text and some augmentation with regards to scaling. By fine-tuning the parameters also for multi-column documents, additional training data was produced that resulted in a much more robust textline detection model.

Image enhancement

This is an image to image model which input was low quality of an image and label was actually the original image. For this one we did not have any GT so we decreased the quality of documents in SBB and then feed them into model.

Scale classification

This is simply an image classifier which classifies images based on their scales or better to say based on their number of columns.

Heuristic methods

Some heuristic methods are also employed to further improve the model predictions:

  • After border detection, the largest contour is determined by a bounding box and the image cropped to these coordinates.
  • For text region detection, the image is scaled up to make it easier for the model to detect background space between text regions.
  • A minimum area is defined for text regions in relation to the overall image dimensions, so that very small regions that are actually noise can be filtered out.
  • Deskewing is applied on the text region level (due to regions having different degrees of skew) in order to improve the textline segmentation result.
  • After deskewing, a calculation of the pixel distribution on the X-axis allows the separation of textlines (foreground) and background pixels.
  • Finally, using the derived coordinates, bounding boxes are determined for each textline.

Installation

pip install . or

pip install . -e for editable installation

Alternatively, you can also use make with these targets:

make install or

make install-dev for editable installation

Models

In order to run this tool you also need trained models. You can download our pretrained models from qurator-data.de.

Alternatively, running make models will download and extract models to $(PWD)/models_eynollah.

Usage

The basic command-line interface can be called like this:

eynollah \
-i <image file name> \
-o <directory to write output xml or enhanced image> \
-m <directory of models> \
-fl <if true, the tool will perform full layout analysis> \
-ae <if true, the tool will resize and enhance the image and produce the resulting image as output> \
-as <if true, the tool will check whether the document needs rescaling or not> \
-cl <if true, the tool will extract the contours of curved textlines instead of rectangle bounding boxes> \
-si <if a directory is given here, the tool will output image regions inside documents there>

The tool does accept and works better on original images (RGB format) than binarized images.

How to use

First of all, this model makes use of up to 9 trained models which are responsible for different operations like size detection, column classification, image enhancement, page extraction, main layout detection, full layout detection and textline detection. But this does not mean that all 9 models are always required for every document. Based on the document characteristics and parameters specified, different scenarios can be applied.

  • If none of the parameters is set to true, the tool will perform a layout detection of main regions (background, text, images, separators and marginals). An advantage of this tool is that it tries to extract main text regions separately as much as possible.

  • If you set -ae (allow image enhancement) parameter to true, the tool will first check the ppi (pixel-per-inch) of the image and when it is less than 300, the tool will resize it and only then image enhancement will occur. Image enhancement can also take place without this option, but by setting this option to true, the layout xml data (e.g. coordinates) will be based on the resized and enhanced image instead of the original image.

  • For some documents, while the quality is good, their scale is extremly large and the performance of tool decreases. In such cases you can set -as (allow scaling) to true. With this option enabled, the tool will try to rescale the image and only then the layout detection process will begin.

  • If you care about drop capitals (initials) and headings, you can set -fl (full layout) to true. With this setting, the tool can currently distinguish 7 document layout classes/elements.

  • In cases where the document includes curved headers or curved lines, rectangular bounding boxes for textlines will not be a great option. In such cases it is strongly recommended to set the flag -cl (curved lines) to true to find countours of curved lines instead of rectangular boundinx boxes. Be advised that enabling this option increases the processing time of the tool.

  • To crop and save image regions inside the document, set the parameter -si (save images) to true and provide a directory path to store the extracted images.

  • This tool is actively being developed. If problems occur or the performance does not meet your expectations, we welcome your feedback via issues.