sajaras / lekha_ocr_ver2.0

Lekha OCR (version 2) is Linux Desktop Application which converts your Malayalam document images to editable Malayalam text with more than 95% accuracy

Lekha Ocr version 2.0

Description

Lekha Ocr version 2.0 Converts your malayalam documents and images to editable malayalam text.It's designed to be easy,fast and simple to use.it has add-on features like scanning,croping,rotating and skew correction of images which help you to do things easy.

In other words, let the machine do most of the malayalam typing work for you.

Screenshots

Main Window

Main Window after convertion

Settings Window

Layout Analyzed Image

Crop

Main features

• Scan
• Import image from disk
• Image layout analysis and feature to correct the result : draw new box by selecting and then press enter to confirm ,delete the incorrect layout analysed box by clicking on respective box.
• Malayalam ocr (lekha_ocrv2)
• Image Skew correction
• Rotate image
• Crop image : crop out unwanted parts of imported/scanned image using crop functionality.
• Set Workspace Directory
• Save Output malayalam text to text file.

Installation : Debian/Ubuntu

• git clone https://gitlab.com/space-kerala/LEKHA_OCR_VER2.0.git (or Download zip and extract)
• Navigate to the cloned or extracted directory
• sudo apt-get update
• sudo bash lekha2-install.sh
• installation done
• Open the application by searching lekha ocr ver 2 in the dash
• Set workspace directory under edit-> preference before any scan operation.

To Uninstall

• sudo bash uninstall.sh

Details

It mainly uses:

• Pyinsane2: To scan the pages
• OPENCV: for image manipulation and calculations
• Lekha_ocrv2(OCR)
• GTK+: For the user interface
• PYGOBJECT: python binding for Gtk+
• Matplotlib: Selection and for layout analyzed output correction features
• Pillow

Creators

Team under Arun M

• Sachin Gracious
• Sajaras k
• Yadhukrishnan K

Contributors

• Arun M helped in project management and technical assistance.
• Ambily Sreekumar,contributed in building data set for training.
• Jithin thankachan:contributed in training tool and helped in documentation.
• Arun Joseph:contributed most of the engine initial developments of Lekha ocr 1.0
• Balagopal Unnikrishnan : Contributed in preparing XML label for training and helped in documentation ofLekha ocr 1.0
• Rijoy V : Contributed in initial research of Lekha OCR ver 1.0

Supporters

This project is developed by SPACE-KERALA in association with ICFOSS

Licence

GPLv3 only. See License.md

Development

All the information can be found on (https://github.com/space-kerala)