Open Access Open Access  Restricted Access Subscription Access

PKIT: Printed Kashmiri Image Text Recognition Using Deep Learning


Affiliations
1 Department of Computer Science, Punjabi University, Patiala,147002, Punjab, India., India
2 Department of Computer Science, Punjabi University, Patiala, 147002, Punjab, India., India
3 Department of Computer Science, Islamic University of Science and Technology, Awantipora, 192122,Jammu and Kashmir, India
 

Optical Character Recognition, often known as OCR, is a method that turns scanned documents, images of text, and PDFs into text documents, that can be edited and searched on a computer. OCR software analyzes a scanned image of text and turns it into machine-encoded text by identifying the characters in the image and transforming them into a digital format. Acknowledging the relevance of optical character recognition (OCR) in the actual world, a multitude of approaches have evolved both for Western and Asian languages. Kashmiri is mostly spoken in the Kashmir Valley, which is located in Jammu and Kashmir India. In spite of the significant amount of effort that has been done into recognizing Indian scripts such as Devanagari, Bengali, Urdu, and Punjabi, no such effort has been made to recognize Kashmiri script. In addition, several benchmark corpora for other Perso Arabic scripts, such as Urdu, Arabic, and Pashto, have been developed for the purpose of training and assessing various OCR systems. Notably, there is currently no OCR corpus for Kashmiri script that can be utilized to train and evaluate deep neural networks for the development of Kashmir OCR. To that purpose, we have proposed a Kashmiri corpus Printed Kashmiri Image Text (PKIT)consisting of 120000 line, and 523000-word level printed text images respectively, well suited for use in deep learning techniques. Additionally, we used the proposed dataset for training different state of art deep learning approaches thereby obtaining the Word Error Rate (WER) and Character Error Rate ((CER)of 5.62% on average.

Keywords

Dataset Generation, Deep Learning, Kashmiri OCR, Optical Character Recognition, Printed Kashmiri Text Recognition.
User
Notifications
Font Size

  • Davis, R., Lyall, J.: Recognition of handwritten characters — a review. Image and Vision Computing 4(4), 208–218 (1986). https://doi.org/10. 1016/0262-8856(86)90048-X
  • Plamondon, R., Srihari, S.N.: On-line and off-line handwriting recognition: A comprehensive survey. IEEE Trans. Pattern Anal. Mach. Intell. 22, 63–84 (2000)
  • Isheawy, N.A.M., Hasan, H.: Optical character recognition (ocr) system. IOSR Journal of Computer Engineering (IOSR-JCE), e-ISSN, 2278–0661 (2015)
  • Gon , calves, G.R., Diniz, M.A., Laroca, R., Menotti, D., Schwartz, W.R.: Real-time automatic license plate recognition through deep multi-task networks. In: 2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), pp. 110–117 (2018). IEEE
  • Qadri, M.T., Asif, M.: Automatic number plate recognition system for vehicle identification using optical character recognition. In: 2009 International Conference on Education Technology and Computer, pp. 335–338 (2009). IEEE
  • Fujisawa, H.: Forty years of research in character and document recognition—an industrial perspective. Pattern Recognition 41(8), 2435–2446 (2008)
  • Mohamed, M., Gader, P.: Handwritten word recognition using segmentation-free hidden markov modeling and segmentation-based dynamic programming techniques. IEEE transactions on pattern analysis and machine intelligence 18(5), 548–554 (1996)
  • Tappert, C.C., Suen, C.Y., Wakahara, T.: The state of the art in online handwriting recognition. IEEE Transactions on pattern analysis and machine intelligence 12(8), 787–808 (1990)
  • Breuel, T.M., Ul-Hasan, A., Al-Azawi, M.A., Shafait, F.: Highperformance ocr for printed english and fraktur using lstm networks. In: 2013 12th International Conference on Document Analysis and Recognition, pp. 683–687 (2013). IEEE
  • Cesar, M., Shinghal, R.: An algorithm for segmenting handwritten postal codes. International journal of man-machine studies 33(1), 63–80 (1990)
  • Cheriet, M., Huang, Y.S., Suen, C.Y.: Background region-based algorithm for the segmentation of connected digits. In: 11th IAPR International Conference on Pattern Recognition. Vol. II. Conference B: Pattern Recognition Methodology and Systems, vol. 1, pp. 619–620 (1992). IEEE Computer Society
  • Doetsch, P., Kozielski, M., Ney, H.: Fast and robust training of recurrentneural networks for offline handwriting recognition. In: 2014 14th International Conference on Frontiers in Handwriting Recognition, pp. 279–284 (2014). IEEE
  • Natarajan, P., Bazzi, I., Lu, Z., Makhoul, J., Scwhartz, R.: Robust ocr of degraded documents. In: Proceedings of the Fifth International Conference on Document Analysis and Recognition. ICDAR’99 (Cat. No. PR00318), pp. 357–361 (1999). IEEE
  • Graves, A., Schmidhuber, J.: Offline handwriting recognition with multidimensional recurrent neural networks. Advances in neural information processing systems 21 (2008)
  • Ashwin, T., Sastry, P.: A font and size-independent ocr system for printed kannada documents using support vector machines. Sadhana 27(1), 35–58 (2002)
  • Bashir, M., Goyal, V., Giri, K.J.: Challenges in recognition of Kashmiri script. In: Singh, P.K., Singh, Y., Kolekar, M.H., Kar, A.K., Gon ,calves, P.J.S. (eds.) Recent Innovations in Computing, pp. 33–43. Springer, Singapore (2022)
  • Bashir, R., Quadri, S.M.K.: Identification of kashmiri script in a bilingual document image. 2013 IEEE Second International Conference on Image Information Processing (ICIIP-2013), 575–579 (2013)
  • Graves, A., Schmidhuber, J.: Offline arabic handwriting recognition with multidimensional recurrent neural networks, pp. 545–552 (2008). https: //doi.org/10.1007/978-1-4471-4072-6 12
  • Pal, A.: Bengali handwritten numeric character recognition using denoising autoencoders. In: 2015 IEEE International Conference on Engineering and Technology (ICETECH), pp. 1–6 (2015). IEEE
  • Ahmad, I., Wang, X., Li, R., Rasheed, S.: Offline urdu nastaleeq optical character recognition based on stacked denoising autoencoder. China Communications 14(1), 146–157 (2017)
  • Naz, S., Umar, A.I., Ahmad, R., Siddiqi, I., Ahmed, S.B., Razzak, M.I., Shafait, F.: Urdu nastaliq recognition using convolutional–recursive deep learning. Neurocomputing 243, 80–87 (2017)
  • Addakiri, K., Bahaj, M.: Article: On-line handwritten arabic character recognition using artificial neural network. International Journal of Computer Applications 55(13), 42–46 (2012). Full text available
  • Osman, H., Zaghw, K., Hazem, M., Elsehely, S.: An efficient languageindependent multi-font OCR for arabic script. CoRR abs/2009.09115 (2020) 2009.09115
  • Fasha, M., Hammo, B.H., Obeid, N., Widian, J.: A hybrid deep learning model for arabic text recognition. ArXiv abs/2009.01987 (2020)
  • Altwaijry, N., Al-Turaiki, I.: Arabic handwriting recognition system using convolutional neural network. Neural Computing and Applications 33 (2021). https://doi.org/10.1007/s00521- 020-05070-8
  • Mushtaq, M.M.M.K.M.K. Faisel, Sing, S.: Urdudeepnet: offline handwritten urdu character recognition using deep neural network. Neural Computing and Applications 33 (2021)
  • Qurat-ul-Ain Akram, F.A.S.U.S.H.S.S. Anneta Niazi: A comprehensive image dataset of urdu nastalique document images, 81–88 (2016)
  • Shi, B., Bai, X., Yao, C.: An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition. IEEE transactions on pattern analysis and machine intelligence 39(11), 2298–2304 (2016)
  • Simonyan, K., Zisserman, A.: Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv (2014). https://doi.org/10.48550/ ARXIV.1409.1556. https://arxiv.org/abs/1409.1556
  • Wei, T.C., Sheikh, U., Ab Rahman, A.A.-H.: Improved optical character recognition with deep neural network. In: 2018 IEEE 14th International Colloquium on Signal Processing & Its Applications (CSPA), pp. 245–249 (2018). IEEE
  • Paul, D., Chaudhuri, B.B.: A BLSTM network for printed bengali OCR system with high accuracy. CoRR abs/1908.08674 (2019) 1908.08674
  • Levenshtein, V.I., et al.: Binary codes capable of correcting deletions, insertions, and reversals. In: Soviet Physics Doklady, vol. 10, pp. 707–710 (1966). Soviet Union

Abstract Views: 185

PDF Views: 0




  • PKIT: Printed Kashmiri Image Text Recognition Using Deep Learning

Abstract Views: 185  |  PDF Views: 0

Authors

Maajid Bashir
Department of Computer Science, Punjabi University, Patiala,147002, Punjab, India., India
Vishal Goyal
Department of Computer Science, Punjabi University, Patiala, 147002, Punjab, India., India
Kaiser J.Giri
Department of Computer Science, Islamic University of Science and Technology, Awantipora, 192122,Jammu and Kashmir, India

Abstract


Optical Character Recognition, often known as OCR, is a method that turns scanned documents, images of text, and PDFs into text documents, that can be edited and searched on a computer. OCR software analyzes a scanned image of text and turns it into machine-encoded text by identifying the characters in the image and transforming them into a digital format. Acknowledging the relevance of optical character recognition (OCR) in the actual world, a multitude of approaches have evolved both for Western and Asian languages. Kashmiri is mostly spoken in the Kashmir Valley, which is located in Jammu and Kashmir India. In spite of the significant amount of effort that has been done into recognizing Indian scripts such as Devanagari, Bengali, Urdu, and Punjabi, no such effort has been made to recognize Kashmiri script. In addition, several benchmark corpora for other Perso Arabic scripts, such as Urdu, Arabic, and Pashto, have been developed for the purpose of training and assessing various OCR systems. Notably, there is currently no OCR corpus for Kashmiri script that can be utilized to train and evaluate deep neural networks for the development of Kashmir OCR. To that purpose, we have proposed a Kashmiri corpus Printed Kashmiri Image Text (PKIT)consisting of 120000 line, and 523000-word level printed text images respectively, well suited for use in deep learning techniques. Additionally, we used the proposed dataset for training different state of art deep learning approaches thereby obtaining the Word Error Rate (WER) and Character Error Rate ((CER)of 5.62% on average.

Keywords


Dataset Generation, Deep Learning, Kashmiri OCR, Optical Character Recognition, Printed Kashmiri Text Recognition.

References