# GraDeT-HTR: A Resource-Efficient Bengali Handwritten Text Recognition System utilizing Grapheme-based Tokenizer and Decoder-only Transformer

arXiv: https://arxiv.org/abs/2509.18081 · HTML: https://mlsanity.com/paper/2509.18081/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 39 references that are in the map.
- First posted on arXiv 22 Sep 2025.

## Builds on (most cited first, 20 of 20)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [You Only Look Once: Unified, Real-Time Object Detection](https://mlsanity.com/paper/1506.02640/) (2015)
- [Neural Machine Translation of Rare Words with Subword Units](https://mlsanity.com/paper/1508.07909/) (2015)
- [An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition](https://mlsanity.com/paper/1507.05717/) (2015)
- [Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition](https://mlsanity.com/paper/2103.06495/) (2021)
- [NRTR: A No-Recurrence Sequence-to-Sequence Model For Scene Text Recognition](https://mlsanity.com/paper/1806.00926/) (2018)
- [What If We Only Use Real Datasets for Scene Text Recognition? Toward Scene Text Recognition With Fewer Labels](https://mlsanity.com/paper/2103.04400/) (2021)
- [On Recognizing Texts of Arbitrary Shapes with 2D Self-Attention](https://mlsanity.com/paper/1910.04396/) (2019)
- [Evaluating Sequence-to-Sequence Models for Handwritten Text Recognition](https://mlsanity.com/paper/1903.07377/) (2019)
- [Handwritten Optical Character Recognition (OCR): A Comprehensive Systematic Literature Review (SLR)](https://mlsanity.com/paper/2001.00139/) (2020)
- [Not Low-Resource Anymore: Aligner Ensembling, Batch Filtering, and New Datasets for Bengali-English Machine Translation](https://mlsanity.com/paper/2009.09359/) (2020)
- [MaskOCR: Text Recognition with Masked Encoder-Decoder Pretraining](https://mlsanity.com/paper/2206.00311/) (2022)
- [A Large Multi-Target Dataset of Common Bengali Handwritten Graphemes](https://mlsanity.com/paper/2010.00170/) (2020)
- [BanglaWriting: A multi-purpose offline Bangla handwriting dataset](https://mlsanity.com/paper/2011.07499/) (2020)
- [End-to-End Optical Character Recognition for Bengali Handwritten Words](https://mlsanity.com/paper/2105.04020/) (2021)
- [LILA-BOTI : Leveraging Isolated Letter Accumulations By Ordering Teacher Insights for Bangla Handwriting Recognition](https://mlsanity.com/paper/2205.11420/) (2022)
- [BN-HTRd: A Benchmark Dataset for Document Level Offline Bangla Handwritten Text Recognition (HTR) and Line Segmentation](https://mlsanity.com/paper/2206.08977/) (2022)
- [Spanish TrOCR: Leveraging Transfer Learning for Language Adaptation](https://mlsanity.com/paper/2407.06950/) (2024)
- [BN-DRISHTI: Bangla Document Recognition through Instance-level Segmentation of Handwritten Text Images](https://mlsanity.com/paper/2306.09351/) (2023)

## Built on by (most cited first, 1 of 1)

- [grapheme-kit: Grapheme-Level Metrics and Text Processing for Multilingual NLP](https://mlsanity.com/paper/2607.22456/) (2026)

## Abstract (arXiv)

Despite Bengali being the sixth most spoken language in the world, handwritten text recognition (HTR) systems for Bengali remain severely underdeveloped. The complexity of Bengali script--featuring conjuncts, diacritics, and highly variable handwriting styles--combined with a scarcity of annotated datasets makes this task particularly challenging. We present GraDeT-HTR, a resource-efficient Bengali handwritten text recognition system based on a Grapheme-aware Decoder-only Transformer architecture. To address the unique challenges of Bengali script, we augment the performance of a decoder-only transformer by integrating a grapheme-based tokenizer and demonstrate that it significantly improves recognition accuracy compared to conventional subword tokenizers. Our model is pretrained on large-scale synthetic data and fine-tuned on real human-annotated samples, achieving state-of-the-art performance on multiple benchmark datasets.