toolri 1.0.1

A tool for extracting, labeling and linking entities in document images for Information Extraction tasks.

ToolRI

ToolRI was created to simplify and standardize the creation of samples for the task of Information Extraction in document images. The tool allows text extraction by OCR, the creation of document entities and their labeling and linking. The project was created purely with Python and can be run on any desktop platform. The graphical user interface is implemented thanks to the amazing CustomTkinter library.

Instalation

PyPi

Install the ToolRI package with pip:

pip install toolri

Source

Clone the ToolRI repository with:

git clone https://github.com/Victorgonl/ToolRI

And install using pip:

pip install ./ToolRI/

Standalone

Download

You can download a portable binary of the tool to start using right away. Download and run a version on the releases page of the ToolRI repository.

Build

To build the standalone version of ToolRI into a portable binary, clone the repository:

git clone https://github.com/Victorgonl/ToolRI

Change current directory to ./ToolRI:

cd ./ToolRI

Install all the dependencies found on requirements.txt:

pip install -r requirements.txt

And run the script toolri_build.py:

python3 toolri_build.py

The binary will be available on dist folder.

Documentation

Under construction. :construction:

Tesseract OCR

To be able to use the OCR function in ToolRI, Tesseract OCR must be installed separately.

For now, OCR is configured for English and Portuguese languages only, but it will be updated soon for all languages available. :construction:

Debian based

Use the command:

sudo apt-get install tesseract-ocr tesseract-ocr-eng tesseract-ocr-por

Windows

• Download and run the installer available at https://github.com/UB-Mannheim/tesseract/wiki.

• Make sure to install Tesseract on C:\Program Files\Tesseract-OCR\ (the default directory) due to a predefined configuration in current ToolRI version.

Usage

ToolRI was developed and used to create the UFLA-FORMS dataset. Download the dataset to try the tool on the available samples or create a new metadata for any document image available.

Example

import toolri

image = toolri.load_image("document_image.jpg")

labels = [
    toolri.ToolRILabel(name="QUESTION", color="#004B80", links=["ANSWER"], is_visible=True),
    toolri.ToolRILabel(name="ANSWER", color="#00943E", links=[], is_visible=True)
]

data = toolri.toolri(image=image, data=data, labels=labels)

toolri.draw_data_on_image(image=image, data=data, labels=labels).show()