HDevChaRNet: A deep learning-based model for recognizing offline handwritten devanagari characters
Abstract
Optical character recognition (OCR) converts text images into machine-readable text. Due to the non-availability of several standard datasets of Devanagari characters, researchers have used many techniques for developing an OCR system with varying recognition rates using their own created datasets. The main objective of our proposed study is to improve the recognition rate by analyzing the effect of using batch normalization (BN) instead of dropout in convolutional neural network (CNN) architecture. So, a CNN-based model HDevChaRNet (Handwritten Devanagari Character Recognition Network) is proposed in this study for same to recognize offline handwritten Devanagari characters using a dataset named Devanagari handwritten character dataset (DHCD). DHCD comprises a total of 46 classes of characters, out of which 36 are consonants, and 10 are numerals. The proposed models based on convolutional neural network (CNN) with BN for recognizing the Devanagari characters showed an improved accuracy of 98.75%, 99.70%, and 99.17% for 36, 10, and 46 classes, respectively.
Keywords
Full Text:
PDFReferences
1. Bal A, Saha R. An improved method for handwritten document analysis using segmentation, baseline recognition and writing pressure detection. Procedia Computer Science 2016; 93: 403–415. doi: 10.1016/j.procs.2016.07.227
2. Bozinovic RM, Srihari SN. Offline cursive script word recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence 1989; 11(1): 68–83. doi: 10.1109/34.23114
3. Qureshi F, Rajput A, Mujtaba G, Fatima N. A novel offline handwritten text recognition technique to convert ruled-line text into digital text through deep neural networks. Multimedia Tools and Applications 2022; 81(13): 18223–18249. doi: 10.1007/s11042-022-12097-7
4. Puri S, Singh SP. An efficient devanagari character classification in printed and handwritten documents using SVM. Procedia Computer Science 2019; 152: 111–121. doi: 10.1016/j.procs.2019.05.033
5. Acharya S, Pant AK, Gyawali PK. Deep learning based large scale handwritten devanagari character recognition. In: Proceedings of 2015 9th International Conference on Software, Knowledge, Information Management and Applications (SKIMA); 15–17 December 2015; Kathmandu, Nepal.
6. Indian A, Bhatia K. A survey of offline handwritten hindi character recognition. In: Proceedings of 2017 3rd International Conference on Advances in Computing, Communication & Automation (ICACCA); 15–16 September 2017; Dehradun, India. pp. 1–6.
7. Bhalerao M, Bonde S, Nandedkar A, Pilawan S. Combined classifier approach for offline handwritten Devanagari character recognition using multiple Features. In: Computational Vision and Bio Inspired Computing, 1st ed. Springer; 2018. pp. 45–54.
8. Ghosh R, Panda C, Kumar P. Handwritten text recognition in bank cheques. In: Proceedings of 2018 Conference on Information and Communication Technology (CICT’18); 26–28 October 2018; Jabalpur, India.
9. Bhatia K, Indian A. Offline handwritten hindi ‘SWARs’ recognition using a novel wave based feature extraction method. International Journal of Computer Science Issues 2017; 14(4): 8–14. doi: 10.20943/01201704.814
10. Puri S, Singh SP. Toward recognition and classification of hindi handwritten document image. In: Ambient Communications and Computer Systems, 1st ed. Springer; 2019. pp. 497–507.
11. Rastogi N, Dutta M, Indian A. Offline handwritten numerals recognition using combinational feature extraction approach. International Journal of Innovative Research in Computer and Communication Engineering 2017; 5(4): 7844–7851.
12. Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems 2012; 25(2): 84–90. doi: 10.1145/3065386
13. Dokare I, Gadge S, Kharde K, et al. Recognition of handwritten devanagari character using convolutional neural network. In: Proceedings of 2021 3rd International Conference on Signal Processing and Communication (ICPSC); 13–14 May 2021; Tamil Nadu, India.
14. Bisht M, Gupta R. Offline handwritten devanagari modified character recognition using convolutional neural network. Sadhana 2021; 46(1). doi:10.1007/s12046-020-01532-w
15. Roy RK, Pal U, Roy K, Kimura F. A system for recognition of destination address in postal documents of India. Malaysian Journal of Computer Science 2020; 3(3): 202–216. doi: 10.22452/mjcs
16. Sharma S, Gupta S, Gupta D, et al. Recognition of gurumukhi handwritten city names using deep learning and cloud computing. Scientific Programming 2022; 2022. doi: 10.1155/2022/5945117
17. Roy RK, Mukherjee H, Roy K, Pal U. CNN based recognition of handwritten multilingual city names. Multimedia Tools and Applications 2022; 81(8): 11501–11517. doi: 10.1007/s11042-022-12193-8
18. Aneja N, Aneja S. Transfer learning using CNN for handwritten devanagari character recognition. In: Proceedings of 2019 1st International Conference on Advances in Information Technology (ICAIT); 25–27 July 2019; Chikmagalur, India.
19. Bjorck J, Gomes C, Selman B, Weinberger KQ. Understanding batch normalization. In: Proceedings of 32nd Conference on Neural Information Processing Systems (NeurIPS 2018); 2–8 December 2018; Montréal, Canada.
20. Garbin C, Zhu X, Marques O. Dropout vs. batch normalization: An empirical study of their impact to deep learning. Multimedia Tools and Applications 2020; 79(19–20): 12777–12815. doi: 10.1007/s11042-019-08453-9
21. Li X, Chen S, Hu X, Yang J. Understanding the disharmony between dropout and batch normalization by variance shift. In: Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 15–20 June 2019; Long Beach CA, USA.
22. Jangid M, Srivastava S. Handwritten devanagari character recognition using layer-wise training of deep convolutional neural networks and adaptive gradient methods. Journal Imaging 2018; 4(2): 41. doi: 10.3390/jimaging4020041
23. Deore SP, Pravin A. Devanagari handwritten character recognition using fine-tuned deep convolutional neural network on trivial dataset. Sadhana 2020; 45(1): 243. doi: 10.1007/s12046-020-01484-1
24. Mhapsekar M, Mhapsekar P, Mhatre A, Sawant V. Implementation of residual network (ResNet) for devanagari handwritten character recognition. In: Vasudevan H, Michalas A, Shekokar N, et al. (editors). Advanced Computing Technologies and Applications. Springer, Singapore; 2020.
25. Gurav Y, Bhagat P, Jadhav R, Sinha S. Devanagari handwritten character recognition using convolutional neural networks. In: Proceedings of 2020 International Conference on Electrical, Communication, and Computer Engineering (ICECCE); 12–13 June 2020; Istanbul, Turkey. pp. 1–6.
26. Manocha SK, Tewari P. Devanagari handwritten character recognition using CNN as feature extractor. In: Proceedings of 2021 International Conference on Smart Generation Computing, Communication and Networking (SMART GENCON); 29–30 October 2021; Pune, India. pp. 1–5.
27. Mishra M, Choudhury T, Sarkar T. Devanagari handwritten character recognition. 2021 IEEE India Council International Subsections Conference (INDISCON) 2021; 1–6. doi: 10.1109/INDISCON53343.2021.9582192
28. Pande SM, Jha BK. Character recognition system for devanagari script using machine learning approach. In: Proceedings of 2021 5th International Conference on Computing Methodologies and Communication (ICCMC); 8–10 April 2021; Erode, India. pp. 899–903.
29. Sachdeva J, Mittal S. Handwritten offline devanagari compound character recognition using machine learning. In: Proceedings of ACI’21: Workshop on Advances in Computational Intelligence at ISIC 2021; 25–27 February 2021; Delhi, India.
30. Sousa Neto AF, Bezerra BLD, Toselli AH, Lima EB. HTR-Flor: A deep learning system for offline handwritten text recognition. In: Proceedings of 2020 33rd Sibgrapi Conference on Graphics, Patterns and Images (SIBGRAPI); 7–10 November 2020; Porto de Galinhas, Brazil.
31. Manchala SY, Kinthali J, Kotha K, et al. Handwritten text recognition using deep learning with tensorflow. International Journal of Engineering Research & Technology (IJERT) 2020; 9(5). doi: 10.17577/IJERTV9IS050534
32. Sree A, Chennamsetti V, Maliakal DT, et al. Handwriting recognition using CNN and RNN. Journal of Chengdu University of Technology (Science and Technology Edition) 2021; 26(7).
33. Gupta N, Liu W. Line segmentation from unconstrained handwritten text images using adaptive approach. Computer Science 2021. doi: 10.48550/arXiv.2104.08777
34. Wang Y, Xiao W, Li S. Offline handwritten text recognition using deep learning: A review. Journal of Physics: Conference Series 2021; 1848(1): 012015. doi: 10.1088/1742-6596/1848/1/012015
35. Mondal R, Malakar S, Smith EHB, Sarkar R. Handwritten English word recognition using a deep learning based object detection architecture. Multimedia Tools and Applications 2022; 81(1): 975–1000. doi: 10.1007/s11042-021-11425-7
36. Kumari L, Singh S, Rathore VVS, Sharma A. A comprehensive handwritten paragraph text recognition system: LexiconNet. arXiv 2022; arXiv:2205.11018. doi: 10.48550/arXiv.2205.11018
37. AlJarrah MN, Zyout MM, Duwairi R. Arabic handwritten characters recognition using convolutional neural network. In: Proceedings of 2021 12th International Conference on Information and Communication Systems (ICICS); 24–26 May 2021; Valencia, Spain, pp. 182–188.
38. Alkhateeb JH, Turani A, Abuhamdah A, et al. An effective deep learning approach for improving off-line arabic handwritten character recognition. International Journal of Software Engineering and Computer Systems 2021; 6(2): 104–112. doi: 10.15282/ijsecs.6.2.2020.7.0076
39. Nayef BH, Abdullah SNHS, Sulaiman R, Alyasseri ZAA. Optimized leaky ReLU for handwritten arabic character recognition using convolution neural networks. Multimedia Tools and Applications 2022; 81: 2065–2094. doi: 10.1007/s11042-021-11593-6
40. Yamashita R, Nishio M, Do RK, Togashi K. Convolutional neural networks: An overview and application in radiology. Insights into Imaging 2018; 9(4): 611–629. doi: 10.1007/s13244-018-0639-9
41. Guha R, Das N, Kundu M, et al. DevNet: An efficient CNN architecture for handwritten devanagari character recognition. International Journal of Pattern Recognition and Artificial Intelligence 2019; 34(12): 20052009. doi: 10.1142/S0218001420520096
42. Romanuke V. Appropriate number and allocation of RELU in convolutional neural network. Research Bulletin of the National Technical University of Ukraine Kyiv Politechnic Institute 2017; 1: 69–78. doi: 10.20535/1810-0546.2017.1.88156
43. Nwankpa C, Ijomah W, Gachagan A, Marshall S. Activation functions: Comparison of trends in practice and research for deep learning. In: Proceedings of 2nd International Conference on Computational Sciences and Technology; 17–19 December 2020; Jamshoro, Pakistan.
44. Srivastava N, Hinton G, Krizhevsky A, et al. Dropout: A simple way to prevent neural networks from overfitting. Journal of Machine Learning Research 2014; 15(1): 1929–1958. doi: 10.5555/2627435.2670313
45. Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In: Proceedings of the 32nd International Conference on International Conference on Machine Learning; 6–11 July 2015; Lille, France. pp. 448–456.
DOI: https://doi.org/10.32629/jai.v6i2.679
Refbacks
Copyright (c) 2023 Bharati Yadav, Ajay Indian, Gaurav Meena
License URL: https://creativecommons.org/licenses/by-nc/4.0