@inproceedings{dubey-etal-2025-multi,
title = "Multi-Feature Graph Convolution Network for {H}indi {OCR} Verification",
author = "Dubey, Shikhar and
Mittal, Krish and
Behera, Sourava Kumar and
Ravikiran, Manikandan and
Kumar, Nitin and
Shigwan, Saurabh and
Saluja, Rohit",
editor = "Bhattacharya, Arnab and
Goyal, Pawan and
Ghosh, Saptarshi and
Ghosh, Kripabandhu",
booktitle = "Proceedings of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA 2025)",
month = dec,
year = "2025",
address = "Mumbai, India",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.bhasha-1.1/",
pages = "1--10",
ISBN = "979-8-89176-313-5",
abstract = "This paper presents a novel Graph Convolutional Network (GCN) based framework for verifying OCR predictions on real Hindi document images, specifically addressing the challenges of complex conjuncts and character segmentation. Our approach first segments Hindi characters in real book images at different levels of granularity, while also synthetically generating word images from OCR predictions. Both real and synthetic images are processed through ResNet-50 to extract feature representations, which are then segmented using multiple patching strategies (uniform, akshara, random, and letter patches). The bounding boxes created using segmentation masks are scaled proportionally to the feature space while extracting features for GCN. We construct a line graph where each node represents a real-synthetic character pair (in feature space). Each node of the line graph captures semantic and geometric features including i) cross-entropy between original and synthetic features, ii) Hu moments difference for shape properties, and iii) and pixel count difference for size variation. The GCN with three convolutional layers (and ELU activation) processes these graph-structured features to verify the correctness of OCR predictions. Experimental evaluation on 1000 images from diverse Hindi books demonstrates the effectiveness of our graph-based verification approach in detecting OCR errors, particularly for challenging conjunct characters where traditional methods struggle."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dubey-etal-2025-multi">
<titleInfo>
<title>Multi-Feature Graph Convolution Network for Hindi OCR Verification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shikhar</namePart>
<namePart type="family">Dubey</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Krish</namePart>
<namePart type="family">Mittal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sourava</namePart>
<namePart type="given">Kumar</namePart>
<namePart type="family">Behera</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Manikandan</namePart>
<namePart type="family">Ravikiran</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nitin</namePart>
<namePart type="family">Kumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Saurabh</namePart>
<namePart type="family">Shigwan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rohit</namePart>
<namePart type="family">Saluja</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA 2025)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Arnab</namePart>
<namePart type="family">Bhattacharya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pawan</namePart>
<namePart type="family">Goyal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Saptarshi</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kripabandhu</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mumbai, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-313-5</identifier>
</relatedItem>
<abstract>This paper presents a novel Graph Convolutional Network (GCN) based framework for verifying OCR predictions on real Hindi document images, specifically addressing the challenges of complex conjuncts and character segmentation. Our approach first segments Hindi characters in real book images at different levels of granularity, while also synthetically generating word images from OCR predictions. Both real and synthetic images are processed through ResNet-50 to extract feature representations, which are then segmented using multiple patching strategies (uniform, akshara, random, and letter patches). The bounding boxes created using segmentation masks are scaled proportionally to the feature space while extracting features for GCN. We construct a line graph where each node represents a real-synthetic character pair (in feature space). Each node of the line graph captures semantic and geometric features including i) cross-entropy between original and synthetic features, ii) Hu moments difference for shape properties, and iii) and pixel count difference for size variation. The GCN with three convolutional layers (and ELU activation) processes these graph-structured features to verify the correctness of OCR predictions. Experimental evaluation on 1000 images from diverse Hindi books demonstrates the effectiveness of our graph-based verification approach in detecting OCR errors, particularly for challenging conjunct characters where traditional methods struggle.</abstract>
<identifier type="citekey">dubey-etal-2025-multi</identifier>
<location>
<url>https://aclanthology.org/2025.bhasha-1.1/</url>
</location>
<part>
<date>2025-12</date>
<extent unit="page">
<start>1</start>
<end>10</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Multi-Feature Graph Convolution Network for Hindi OCR Verification
%A Dubey, Shikhar
%A Mittal, Krish
%A Behera, Sourava Kumar
%A Ravikiran, Manikandan
%A Kumar, Nitin
%A Shigwan, Saurabh
%A Saluja, Rohit
%Y Bhattacharya, Arnab
%Y Goyal, Pawan
%Y Ghosh, Saptarshi
%Y Ghosh, Kripabandhu
%S Proceedings of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA 2025)
%D 2025
%8 December
%I Association for Computational Linguistics
%C Mumbai, India
%@ 979-8-89176-313-5
%F dubey-etal-2025-multi
%X This paper presents a novel Graph Convolutional Network (GCN) based framework for verifying OCR predictions on real Hindi document images, specifically addressing the challenges of complex conjuncts and character segmentation. Our approach first segments Hindi characters in real book images at different levels of granularity, while also synthetically generating word images from OCR predictions. Both real and synthetic images are processed through ResNet-50 to extract feature representations, which are then segmented using multiple patching strategies (uniform, akshara, random, and letter patches). The bounding boxes created using segmentation masks are scaled proportionally to the feature space while extracting features for GCN. We construct a line graph where each node represents a real-synthetic character pair (in feature space). Each node of the line graph captures semantic and geometric features including i) cross-entropy between original and synthetic features, ii) Hu moments difference for shape properties, and iii) and pixel count difference for size variation. The GCN with three convolutional layers (and ELU activation) processes these graph-structured features to verify the correctness of OCR predictions. Experimental evaluation on 1000 images from diverse Hindi books demonstrates the effectiveness of our graph-based verification approach in detecting OCR errors, particularly for challenging conjunct characters where traditional methods struggle.
%U https://aclanthology.org/2025.bhasha-1.1/
%P 1-10
Markdown (Informal)
[Multi-Feature Graph Convolution Network for Hindi OCR Verification](https://aclanthology.org/2025.bhasha-1.1/) (Dubey et al., BHASHA 2025)
ACL
- Shikhar Dubey, Krish Mittal, Sourava Kumar Behera, Manikandan Ravikiran, Nitin Kumar, Saurabh Shigwan, and Rohit Saluja. 2025. Multi-Feature Graph Convolution Network for Hindi OCR Verification. In Proceedings of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA 2025), pages 1–10, Mumbai, India. Association for Computational Linguistics.