Handw_recognition

This repository contains the code of our submission for Accenture Digital Challenge.. Other contributor is Ashutosh Sancheti.

Introduction

The participants were asked to use their skills with image recognition to find an innovative solution based on artificial intelligence able to recognize hand written texts and manuscripts (at least in block letters) in latin alphabet and letters.

Demo Video

Dataset:

• To use test_iam_dataset.ipynb in src/utils/, create credentials.json using credentials.json.example and editing the appropriate field. The username and password can be obtained from http://www.fki.inf.unibe.ch/DBs/iamDB/iLogin/index.php.

Steps to Reproduce

1. Clone the repository to your machine.

git clone https://github.com/limosin/Handw_recognition

2. Setup the virtual environemnt using conda.

conda env create -f environment.yml

3. Download the pretrained weights from this link.
4. Extract to '/models' in the main directory(Make this folder if it does not exists).
5. Now for performing OCR on a 'image_example.jpg', open a terminal in the main directory and enter this command.

python OCR.py -f <image_example>

You can refer to the Demo for more detailed steps.

SClite installation

1. Navigate to src/utils/ and find sctk-2.4.10-20151007-1312Z.tar.bz2.
2. Untar sctk-2.4.10
3. Install sctk-2.4.10 by following sctk-2.4.10/INSTALL
4. Check sctk-2.4.10/bin contains built programs

References

The original prototype was built by Thomas Delteil.