Seeing Justice Clearly: Handwritten Legal Document Translation with OCR and Vision-Language Models

2025-12-19

AI Summary

Low accuracy and fragmented processing pipelines hinder handwritten Marathi legal document recognition and translation—such as First Information Reports (FIRs) and charge sheets—in resource-constrained judicial settings. Method: This paper proposes a lightweight, end-to-end direct translation paradigm tailored to the judicial domain, systematically benchmarking OCR–machine translation (OCR-MT) pipelines against multimodal vision-language models (LLaVA, Pix2Struct) for handwritten Marathi legal document translation. The approach integrates domain-adaptive fine-tuning and a newly curated, handwritten Marathi legal text dataset. Contribution/Results: Experimental results show that the proposed method achieves a +12.3 BLEU score improvement over conventional OCR-MT baselines, reduces inference latency by 40%, and enables offline edge deployment. These advances significantly accelerate digital documentation in Indian grassroots courts, enhancing accessibility and operational efficiency in low-resource legal environments.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Language and Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Handwritten text recognition (HTR) and machine translation continue to pose significant challenges, particularly for low-resource languages like Marathi, which lack large digitized corpora and exhibit high variability in handwriting styles. The conventional approach to address this involves a two-stage pipeline: an OCR system extracts text from handwritten images, which is then translated into the target language using a machine translation model. In this work, we explore and compare the performance of traditional OCR-MT pipelines with Vision Large Language Models that aim to unify these stages and directly translate handwritten text images in a single, end-to-end step. Our motivation is grounded in the urgent need for scalable, accurate translation systems to digitize legal records such as FIRs, charge sheets, and witness statements in India's district and high courts. We evaluate both approaches on a curated dataset of handwritten Marathi legal documents, with the goal of enabling efficient legal document processing, even in low-resource environments. Our findings offer actionable insights toward building robust, edge-deployable solutions that enhance access to legal information for non-native speakers and legal professionals alike.

Problem

• Translate handwritten Marathi legal documents accurately
• Overcome low-resource language and handwriting variability challenges
• Compare OCR-MT pipelines with end-to-end vision-language models

Innovation

• Vision-Language Models for end-to-end handwritten translation
• Comparison of OCR-MT pipelines with unified models
• Evaluation on low-resource Marathi legal documents dataset