hicr-capsnet

Tensorflow 2.0 implementation of "Handwritten Indic Character Recognition using Capsule Networks" [ASPCON 2018]

Initial Setup

1. Setup the environment & install all the dependencies:
   pip install -r requirements.txt
2. Download the datasets Bangla Digits, Telugu Digits, Devanagari Digits and rename as bangla, devanagari, telugu in data directory
3. python data/process_dataset.py
4. If you want to create csv file from images folder then: python data/process_dataset.py --csv

Running

• For starting training & validation:
  python main.py

Acknowledgements

A major thanks to authors of the paper.

Reference

@misc{1901.00166,
Author = {Bodhisatwa Mandal and Suvam Dubey and Swarnendu Ghosh and Ritesh Sarkhel and Nibaran Das},
Title = {Handwritten Indic Character Recognition using Capsule Networks},
Year = {2019},
Eprint = {arXiv:1901.00166},
}