Back to Search
Start Over
AI based offline handwritten text recognition system.
- Source :
- AIP Conference Proceedings; 11/8/2022, Vol. 2481 Issue 1, p1-10, 10p
- Publication Year :
- 2022
-
Abstract
- Despite advances in technology Handwritten text is still considered as a method for correspondence and recording data in everyday life. Offline handwritten text recognition is an infamously challenging problem in Artificial Intelligence. Recent studies show great potential of Neural Networks for handwritten text recognition segmented at word level. In this paper, a prototype is presented for offline handwritten text recognition using deep neural networks. Preprocessing techniques like data augmentation and line-level segmentation enrich the input images and hence benefits the classifier. The proposed approach performs thorough feature engineering on preprocessed input images and propagates them through a three layered architectures comprising of CNN (Convolutional Neural Network), LSTM(Long Short Term Memory) and CTC(Connectionist Temporal Classification). [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 0094243X
- Volume :
- 2481
- Issue :
- 1
- Database :
- Complementary Index
- Journal :
- AIP Conference Proceedings
- Publication Type :
- Conference
- Accession number :
- 160090417
- Full Text :
- https://doi.org/10.1063/5.0104582