BengaliNet: A Low-Cost Novel Convolutional Neural Network for Bengali Handwritten Characters Recognition

As it is the seventh most-spoken language and fifth most-spoken native language in the world, the domain of Bengali handwritten character recognition has fascinated researchers for decades. Although other popular languages i.e., English, Chinese, Hindi, Spanish, etc. have received many contributions in the area of handwritten character recognition, Bengali has not received many noteworthy contributions in this domain because of the complex curvatures and similar writing fashions of Bengali characters. Previously, studies were conducted by using different approaches based on traditional learning, and deep learning. In this research, we proposed a low-cost novel convolutional neural network architecture for the recognition of Bengali characters with only 2.24 to 2.43 million parameters based on the number of output classes. We considered 8 different formations of CMATERdb datasets based on previous studies for the training phase. With experimental analysis, we showed that our proposed system outperformed previous works by a noteworthy margin for all 8 datasets. Moreover, we tested our trained models on other available Bengali characters datasets such as Ekush, BanglaLekha, and NumtaDB datasets. Our proposed architecture achieved 96–99% overall accuracies for these datasets as well. We believe our contributions will be beneficial for developing an automated high-performance recognition tool for Bengali handwritten characters.

[1]  Samiul Alam,et al.  NumtaDB - Assembled Bengali Handwritten Digits , 2018, ArXiv.

[2]  N. Kasthuri,et al.  An efficient recognition system for preserving ancient historical documents of English characters , 2020 .

[3]  Sudipta Roy,et al.  Bengali Handwritten Character Classification Using Transfer Learning on Deep Convolutional Network , 2019, IHCI.

[4]  Benjamin Schrauwen,et al.  Deep content-based music recommendation , 2013, NIPS.

[5]  Nagib Mahfuz,et al.  Convolutional Neural Networks Based Bengali Handwritten Character Recognition , 2020 .

[6]  Andreas Holzinger,et al.  Augmentor: An Image Augmentation Library for Machine Learning , 2017, J. Open Source Softw..

[7]  Changshui Zhang,et al.  Zero-shot Handwritten Chinese Character Recognition with hierarchical decomposition embedding , 2020, Pattern Recognit..

[8]  Subhadip Basu,et al.  Handwritten Bangla Alphabet Recognition using an MLP Based Classifier , 2012, ArXiv.

[9]  S. Imam.  English as a Global Language and the Question of Nation-Building Education in Bangladesh. , 2005 .

[10]  Malayappan Shridhar,et al.  On Recognition of Handwritten Bangla Characters , 2006, ICVGIP.

[11]  Marc Schoenauer,et al.  An artificial immune system for offline isolated handwritten arabic character recognition , 2018, Evol. Syst..

[12]  Mohamed Hamada,et al.  Attention-Based Fully Gated CNN-BGRU for Russian Handwritten Text , 2020, J. Imaging.

[13]  S. M. Salim Reza,et al.  Bangla handwritten character recognition using MobileNet V1 architecture , 2020, Bulletin of Electrical Engineering and Informatics.

[14]  Mahantapas Kundu,et al.  A multi-scale deep quad tree based feature extraction method for the recognition of isolated handwritten characters of popular indic scripts , 2017, Pattern Recognit..

[15]  Sajjad Waheed,et al.  Handwritten Bangla Character Recognition Using Deep Convolutional Neural Network: Comprehensive Analysis on Three Complete Datasets , 2021 .

[16]  Showmik Bhowmik,et al.  Language-invariant novel feature descriptors for handwritten numeral recognition , 2020, The Visual Computer.

[17]  Jae Hyuk Heo,et al.  A Comparative Study on the Perception Performance of Handwriting in Korean and English Using Machine Learning , 2021, 2021 21st ACIS International Winter Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD-Winter).

[18]  Pakpoom Mookdarsanit,et al.  THAIWRITTENNET: THAI HANDWRITTEN SCRIPT RECOGNITION USING DEEP NEURAL NETWORKS , 2020 .

[19]  Rafiqul Islam,et al.  BanglaLekha-Isolated: A multi-purpose comprehensive dataset of Handwritten Bangla Isolated characters , 2017, Data in brief.

[20]  Nafees Mansoor,et al.  Classification of Bangla Compound Characters Using a HOG-CNN Hybrid Model , 2018 .

[21]  Mita Nasipuri,et al.  A multi-objective approach towards cost effective isolated handwritten Bangla character and digit recognition , 2016, Pattern Recognit..

[22]  P. A. Lyakhov,et al.  Application of the residue number system to reduce hardware costs of the convolutional neural network implementation , 2020, Math. Comput. Simul..

[23]  Subhadip Basu,et al.  An Improved Feature Descriptor for Recognition of Handwritten Bangla Alphabet , 2015, ArXiv.

[24]  Syeda Rumnaz Imam.  English as a global language and the question of nation‐building education in Bangladesh , 2005 .

[25]  Ji Gan,et al.  Compressing the CNN architecture for in-air handwritten Chinese character recognition , 2020, Pattern Recognit. Lett..

[26]  Soumen Bag,et al.  A Study on the Effect of CNN-Based Transfer Learning on Handwritten Indic and Mixed Numeral Recognition , 2018, DAR@ICVGIP.

[27]  Mahmoud Badawy,et al.  A new Arabic handwritten character recognition deep learning system (AHCR-DLS) , 2020, Neural Computing and Applications.

[28]  Subhadip Basu,et al.  Handwritten Bangla character recognition using a soft computing paradigm embedded in two pass approach , 2015, Pattern Recognit..

[29]  Shrinivas R. Zanwar,et al.  Handwritten English Character Recognition Using Swarm Intelligence and Neural Network , 2020 .

[30]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[31]  Jishnu Mukhoti,et al.  Handwritten Digit Classification in Bangla and Hindi Using Deep Learning , 2020, Appl. Artif. Intell..

[32]  Mohammed Aarif K.O,et al.  OCR-Nets: Variants of Pre-trained CNN for Urdu Handwritten Character Recognition via Transfer Learning , 2020 .

[33]  SOUMEN BAG,et al.  A survey on optical character recognition for Bangla and Devanagari scripts , 2013, Sadhana.

[34]  Subhadip Basu,et al.  A benchmark image database of isolated Bangla handwritten compound characters , 2014, International Journal on Document Analysis and Recognition (IJDAR).

[35]  Najla Al-Nabhan,et al.  Analysis and comparison of machine learning classifiers and deep neural networks techniques for recognition of Farsi handwritten digits , 2020, The Journal of Supercomputing.

[36]  Prasant Kumar Pattnaik,et al.  A Robust Approach with Text Analytics for Bengali Digit Recognition Using Machine Learning , 2021 .

[37]  Partha Pratim Roy,et al.  Handwritten Bangla character and numeral recognition using convolutional neural network for low-memory GPU , 2019, International Journal of Machine Learning and Cybernetics.

[38]  Chafic Mokbel,et al.  Transcription of Spanish Historical Handwritten Documents with Deep Neural Networks , 2018, J. Imaging.

[39]  Albert Pravin,et al.  Devanagari Handwritten Character Recognition using fine-tuned Deep Convolutional Neural Network on trivial dataset , 2020, Sādhanā.

[40]  Mahantapas Kundu,et al.  Multiobjective optimization for recognition of isolated handwritten Indic scripts , 2019, Pattern Recognit. Lett..

[41]  Syed Akhter Hossain,et al.  Ekush: A Multipurpose and Multitype Comprehensive Database for Online Off-Line Bangla Handwritten Characters , 2018, RTIP2R.