Related Experiment Video
Updated: Mar 10, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.6K
Urdu Nasta'liq text recognition using implicit segmentation based on multi-dimensional long short term memory neural
Saeeda Naz1, Arif Iqbal Umar2, Riaz Ahmed3
1National University of Sciences and Technology (NUST), Islamabad, Pakistan ; Department of Information Technology, Hazara University, Mansehra, Pakistan.
Springerplus
|December 13, 2016
Summary
This study introduces a novel Multi-dimensional Long Short Term Memory (MDLSTM) Recurrent Neural Network for Urdu text recognition. The developed model achieves 98% accuracy for Nasta'liq script, outperforming existing methods.
Area of Science:
- Computer Science
- Artificial Intelligence
- Natural Language Processing
Background:
- Arabic script recognition is challenging due to its complexity.
- Urdu text recognition, specifically, is hindered by the Nasta'liq style's calligraphic nature, including diagonality, cursiveness, context sensitivity, and character overlap.
- Existing Arabic script recognition methods are not directly applicable to Urdu.
Purpose of the Study:
- To develop an effective method for recognizing printed Urdu text-lines in the Nasta'liq writing style.
- To address the specific challenges posed by the Nasta'liq script for automated text recognition.
Main Methods:
- Implementation of Multi-dimensional Long Short Term Memory (MDLSTM) Recurrent Neural Networks.
- Design of a specialized output layer for sequence labeling tailored for Urdu text recognition.
Main Results:
- Achieved a 98% recognition accuracy for unconstrained Urdu Nasta'liq printed text.
- Demonstrated significant performance improvement over state-of-the-art techniques for Urdu text recognition.
Conclusions:
- MDLSTM Recurrent Neural Networks are highly effective for Urdu Nasta'liq text recognition.
- The proposed method offers a substantial advancement in accurately recognizing complex Urdu script, paving the way for improved digital processing of Urdu documents.
