Hybrid Modeling for Effective Text Spotting in Gujarati Language

Riddhi Dave and Sheshang Degadwala

International Journal of Scientific Research in Science and Technology, 2025, vol. 12, issue 3, 411-416

Abstract: This paper presents a novel hybrid modeling approach for effective text spotting specifically tailored to the Gujarati language, achieving a high accuracy of 91.8% while maintaining efficient training time of only 18 minutes. The proposed hybrid model synergistically combines convolutional neural networks (CNN) for feature extraction and transformer-based architectures for contextual understanding, optimizing both recognition accuracy and computational efficiency. Gujarati, with its complex script and unique character shapes, presents challenges such as cursive and ligature forms, which the hybrid framework effectively addresses by leveraging spatial and sequential information jointly. Unlike traditional single-method models that either focus solely on spatial features or sequential patterns, our hybrid approach integrates both aspects to improve robustness against noise, background clutter, and varying text orientations in natural scene images. Experimental results on a custom-compiled dataset of Gujarati text in diverse scenes demonstrate superior performance compared to baseline models, with a notable reduction in false positives and recognition errors. The short training time also makes this method viable for real-world applications requiring quick model updates or deployment on resource-constrained devices. This work contributes a valuable advancement in Indic script OCR research and opens pathways for extending hybrid frameworks to other low-resource languages with complex scripts. The system’s effectiveness in handling multi-style text and mixed backgrounds suggests promising potential for integration into mobile-based text reading applications and automated document processing for Gujarati text.

Keywords: Text spotting; deep learning; scene text detection; optical character recognition (OCR); end-to-end frameworks; machine learning; computer vision (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://ijsrst.com/home/article/view/IJSRST2512356 Abstract page (text/html)
https://ijsrst.com/home/article/download/IJSRST2512356/IJSRST2512356 Full text (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:etm:ijsrst:v12:y2025:i3:id:859

DOI: 10.32628/IJSRST2512356

Access Statistics for this article

More articles in International Journal of Scientific Research in Science and Technology from Technoscience Academy
Bibliographic data for series maintained by Pankaj Sharma ().