Handwritten Bangla Character Recognition Using Deep Convolutional Neural Network: Comprehensive Analysis on Three Comple
Bangla handwritten character recognition is a difficult job compared to other languages due to the morphological complexity of adjacent characters and a wide variety of curvatures in writing styles people have. Another reason for that is the unique presen
- PDF / 540,096 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 75 Downloads / 229 Views
Abstract Bangla handwritten character recognition is a difficult job compared to other languages due to the morphological complexity of adjacent characters and a wide variety of curvatures in writing styles people have. Another reason for that is the unique presence of compound characters. Most of the recent research works conducted in this field standardize Deep Convolutional Neural Network (DCNN) models for delivering the most effective outcomes. This paper proposes a DCNN model to classify all the character classes from three popular databases known as BanglaLekha Isolated, Ekush, and CMATERdb. As for BanglaLekha Isolated, our model achieves 93.446% accuracy on the 50 alphabets category and an overall 91.45% considering the whole dataset. The other two datasets, Ekush and CMATERdb result in 95.05% and 94.17% respectively, where the second one holds 171 classes of compound characters alone and performs 93.259% correctness, which is so far the best for this specific category in this dataset. Keywords Handwritten bangla character · Deep convolutional neural network · Banglalekha · Ekush · CMATERdb · Three full datasets · 28 × 28 images · Bangla compound characters
M. Mashrukh Zayed · S. M. Neyamul Kabir Utsha · S. Waheed (B) Mawlana Bhashani Science and Technology University, Tangail, Dhaka 1902, Bangladesh e-mail: [email protected] URL: https://mbstu.ac.bd/ M. Mashrukh Zayed e-mail: [email protected] URL: https://mbstu.ac.bd/ S. M. Neyamul Kabir Utsha e-mail: [email protected] URL: https://mbstu.ac.bd/ © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 M. S. Kaiser et al. (eds.), Proceedings of International Conference on Trends in Computational and Cognitive Engineering, Advances in Intelligent Systems and Computing 1309, https://doi.org/10.1007/978-981-33-4673-4_7
77
78
M. Mashrukh Zayed et al.
1 Introduction In recent years, handwritten character recognition has emerged as one of the major interests to the computer vision researchers due to its various application potentials. It has many significant usages in our regular life, such as zip code scanning, bank check processing, identifying postal codes, reading national ID numbers, etc. most of which are handwritten documents. This paper focuses on the recognition of Bangla handwritten characters and the measurement of accuracies achieved on three complete datasets available for handwritten Bangla alphabets. Between handwritten and printed Bangla characters, recognizing handwritten characters is more challenging and complicated due to the following reasons: (i) Bangla alphabets have a wide range of characters that are morphologically complex (compound characters), which make the recognition task even more challenging; (ii) Different persons have individual writing styles, so the same character written by different writers vary in shapes, sizes, and curvatures; (iii) The similarities between some characters in shapes further complicates the recognition problem. In this paper, we developed a handwritten character rec
Data Loading...
