It is a machine learning based web application for performing end-to-end full-page handwriting recognition using the Apache MXNet deep learning frameworks on the IAM Dataset. This handwriting recognition system would overcome the use of storing physical documents and reducing the storage of image data. Physical documents such as examination papers would be not taking up unnecessary amount of space but instead being transcribed into digital format. By reducing storage, more spaces could be used for meaningful use and data could be transported easily instead of passing the papers around. With the use of handwriting recognition system, researchers could preserve important documents which are prone to decay to due lifetime use of and physical deterioration. This would be a whole new level of access for researchers to access historical manuscripts in a totally new way. It would enable automatic reading of ancient manuscripts and that may lead to an increase of availability of the content.
Fork the repo and start it on your Python IDE.
Refer requirements.txt
- Flask - The web framework used
- MXNet Handwriting Recognition Models - AWS Handwriting Recognition
This project is licensed under the MIT License - see the LICENSE.md file for details
- My Supervisor Mr Syed Mohd Zahid Bin Syed Zainal Arrifin for his guidance and support
- Family for their support and encouragement
- Deanna Lee for her guidance and help on chinese translation