BornoViT: A Novel Efficient Vision Transformer for Bengali Handwritten Basic Characters Classification

Rafi Hassan Chowdhury; Naimul Haque; Kaniz Fatiha

cs.CVFeb 28, 2026

BornoViT: A Novel Efficient Vision Transformer for Bengali Handwritten Basic Characters Classification

Authors: Rafi Hassan Chowdhury, Naimul Haque, Kaniz Fatiha

Organizations: Department of Computer Science and Engineering, Islamic University of Technology, Gazipur 1704, Bangladesh

Abstract

Handwritten character classification in the Bengali script is a significant challenge due to the complexity and variability of the characters. The models commonly used for classification are often computationally expensive and data-hungry, making them unsuitable for resource-limited languages such as Bengali. In this experiment, we propose a novel, efficient, and lightweight Vision Transformer model that effectively classifies Bengali handwritten basic characters and digits, addressing several shortcomings of traditional methods. The proposed solution utilizes a deep convolutional neural network (DCNN) in a more simplified manner compared to traditional DCNN architectures, with the aim of reducing computational burden. With only 0.65 million parameters, a model size of 0.62 MB, and 0.16 GFLOPs, our model, BornoViT, is significantly lighter than current state-of-the-art models, making it more suitable for resource-limited environments, which is essential for Bengali handwritten character classification. BornoViT was evaluated on the BanglaLekha Isolated dataset, achieving an accuracy of 95.77%, and demonstrating superior efficiency compared to existing state-of-the-art approaches. Furthermore, the model was evaluated on our self-collected dataset, Bornomala, consisting of approximately 222 samples from different age groups, where it achieved an accuracy of 91.51%.

Figures & tables

Fig. 2: Sample of BanglaLekha-Isolated Dataset

Panel of Figure 3 — Fig. 3: Sample of augmentation techniques

Layer	Description	Input Shape	Output Shape	Parameters
PatchEmbedding	Conv2D layer to split image into patches and project them to embedding dimension	(B, 3, 224, 224)	(B, 196, 128)	98,432
Positional Embedding	Learnable positional embedding added to the input patches	(B, 197, 128)	(B, 197, 128)	25,216
CLS Token	Learnable CLS token for classification	(1, 1, 128)	(B, 1, 128)	128
Dropout	Dropout applied to the patch and position embeddings	(B, 197, 128)	(B, 197, 128)	0
Transformer Blocks (x4)	Sequence of 4 Transformer blocks	(B, 197, 128)	(B, 197, 128)	4x132,096 (Table III )
LayerNorm	Normalize embeddings	(B, 197, 128)	(B, 197, 128)	256

TABLE II: Layer-wise Breakdown of Custom Vision Transformer

Sub-layer	Description	Input Shape	Output Shape	Parameters
LayerNorm 1	Layer normalization	(B, 197, 128)	(B, 197, 128)	256
Multi-Head Attention	Self-attention mechanism with 2 attention heads	(B, 197, 128)	(B, 197, 128)	49,152 (QKV), 16,384 (O)
LayerNorm 2	Layer normalization	(B, 197, 128)	(B, 197, 128)	256
MLP	Two linear layers with GELU and dropout	(B, 197, 128)	(B, 197, 128)	33,024 (fc1), 33,024 (fc2)
Total				132,096

TABLE III: Transformer Block Breakdown

Panel of Figure 6 — Fig. 4: GradCam samples of correctly classified inputs of BanglaLekha-Isolated

Panel of Figure 7 — Fig. 5: GradCam samples of correctly classified inputs of Bornomala

Panel of Figure 8 — Fig. 6: Misclassified samples that are visually similar to the predicted class.

Explore similar work

Maximizing Generalization: The Effect of Different Augmentation Techniques on Lightweight Vision Transformer for Bengali Character Classification
Mar 3, 2026Rafi Hassan Chowdhury, Naimul Haque, Kaniz FatihaData Augmentation Optical Character Recognition
Department of Computer Science and Engineering, Islamic University of Technology, Gazipur 1704, Bangladesh
Multi-Head Attention based interaction-aware architecture for Bangla Handwritten Character Recognition: Introducing a Primary Dataset
Apr 8, 2026Mirza Raquib, Asif Pervez Polok, Kedar Nath Biswas +3Optical Character Recognition Hybrid CNN-Transformer Architectures
Department of Computer and Communication Engineering, International Islamic University Chittagong, Bangladesh. · mPower Social Enterprise. · Department of Information and Communication Engineering, Noakhali Science and Technology University, Bangladesh. +2
Confidence-Guided Diffusion Augmentation for Enhanced Bangla Compound Character Recognition
May 11, 2026Md. Sultan Al RayhanSynthetic Data Augmentation Optical Character Recognition
Department of Computer Science and Engineering East West University, Dhaka, Bangladesh

Mar 3, 2026J/K move · Enter open · S save

Maximizing Generalization: The Effect of Different Augmentation Techniques on Lightweight Vision Transformer for Bengali Character Classification

Rafi Hassan Chowdhury, Naimul Haque, Kaniz Fatiha

Department of Computer Science and Engineering, Islamic University of Technology, Gazipur 1704, Bangladesh

Data Augmentation Optical Character Recognition Low-Resource Language Processing

Open paper AskPDF arXiv

Deep learning models have proven to be highly effective in computer vision, with deep convolutional neural networks achieving impressive results across various computer vision tasks. However, these models rely heavily on large datasets to avoid overfitting. When a model learns features with either low or high variance, it can lead to underfitting or overfitting on the training data. Unfortunately, large-scale datasets may not be available in many domains, particularly for resource-limited languages such as Bengali. In this experiment, a series of tests were conducted in the field of image data augmentation as an approach to addressing the limited data problem for Bengali handwritten characters. The study also provides an in-depth analysis of the performance of different augmentation techniques. Data augmentation refers to a set of techniques applied to data to increase its size and diversity, making it more suitable for training deep learning models. The image augmentation techniques evaluated in this study include CLAHE, Random Rotation, Random Affine, Color Jitter, and their combinations. The study further explores the use of augmentation methods with a lightweight model such as EfficientViT. Among the different augmentation strategies, the combination of Random Affine and Color Jitter produced the best accuracy on the Ekush [1] and AIBangla [2] datasets, achieving accuracies of 97.48% and 97.57%, respectively. This combination outperformed all other individual and combined augmentation techniques. Overall, this analysis presents a thorough examination of the impact of image data augmentation in resource-scarce languages, particularly in the context of Bengali handwritten character recognition using lightweight models.

Abstract

Explore similar work

Maximizing Generalization: The Effect of Different Augmentation Techniques on Lightweight Vision Transformer for Bengali Character Classification

Multi-Head Attention based interaction-aware architecture for Bangla Handwritten Character Recognition: Introducing a Primary Dataset

Confidence-Guided Diffusion Augmentation for Enhanced Bangla Compound Character Recognition