Multi-Digit Number Recognition System: Single-Digit CNNs for Multi-Digit Detection and Recognition Using MNIST Dataset

Jehad Ur Rahman; Sara Almarzooqi

Authors

Jehad Ur Rahman University of Engineering and Technology Peshawar
Sara College of Technological Innovation, Zayed University, UAE

Keywords:

MNIST, CNN, Deep Learning, Augmentation, Digit Recognition

Abstract

This research focuses on creating a deep-learning model for identifying multi-digit numbers, which addresses the critical demand for accuracy in real-world applications. The study presents novel approaches to multi-digit recognition, providing a thorough resolution to an unsolved problem in the field of computer vision. In order to improve model generalization, the study makes use of convolutional neural networks (CNNs) that were trained on the MNIST dataset and augmented with rotation and scaling approaches. Multi-digit number prediction is a multi-step process. detection to isolate each digit. Each digit is then clipped and stored separately with its own label. Subsequently, the algorithm predicts the digit for each cropped picture and saves them. This method is repeated for all identified contours, with each predicted digit concatenated to get the final multi-digit prediction. Finally, the projected multi-digit sequence is compared to the ground truth for assessment. The CNN achieves remarkable training and validation accuracies of 99.60% and 99.28%, proving its ability to recognize multi-digit numbers. This study emphasizes the importance of advanced methods in developing deep learning models for multi-digit recognition, which promise enhanced automation and efficiency across a variety of digital technology industries.

References

S. Ahlawat, A. Choudhary, A. Nayyar, S. Singh, and B. Yoon, “Improved Handwritten Digit Recognition Using Convolutional Neural Networks (CNN),” Sensors 2020, Vol. 20, Page 3344, vol. 20, no. 12, p. 3344, Jun. 2020, doi: 10.3390/S20123344.

A. M. Crawford, D. M. Ommen, and A. L. Carriquiry, “A statistical approach to aid examiners in the forensic analysis of handwriting,” J. Forensic Sci., vol. 68, no. 5, pp. 1768–1779, Sep. 2023, doi: 10.1111/1556-4029.15337.

R. A. Hicklin et al., “Accuracy and reliability of forensic handwriting comparisons,” Proc. Natl. Acad. Sci. U. S. A., vol. 119, no. 32, p. e2119944119, Aug. 2022, doi: 10.1073/PNAS.2119944119/SUPPL_FILE/PNAS.2119944119.SD05.XLSX.

A. Sharma and A. Sharma, “A combined static and dynamic feature extraction technique to recognize handwritten digits,” Vietnam J. Comput. Sci. 2015 23, vol. 2, no. 3, pp. 133–142, Jan. 2015, doi: 10.1007/S40595-014-0038-1.

M. Asif, M. Bin Ahmad, S. Mushtaq, K. Masood, T. Mahmood, and A. Ali Nagra, “Long Multi-digit Number Recognition from Images Empowered by Deep Convolutional Neural Networks,” Comput. J., vol. 65, no. 10, pp. 2815–2827, Oct. 2022, doi: 10.1093/COMJNL/BXAB117.

I. J. Goodfellow, Y. Bulatov, J. Ibarz, S. Arnoud, and V. Shet, “Multi-digit Number Recognition from Street View Imagery using Deep Convolutional Neural Networks,” 2nd Int. Conf. Learn. Represent. ICLR 2014 - Conf. Track Proc., Dec. 2013, Accessed: May 25, 2024. [Online]. Available: https://arxiv.org/abs/1312.6082v4

C. L. Tan and A. Jantan, “Digit Recognition Using Neural Networks,” Malaysian J. Comput. Sci., vol. 17, no. 2, pp. 40–54, Dec. 2004, Accessed: May 25, 2024. [Online]. Available: https://ejournal.um.edu.my/index.php/MJCS/article/view/6216

F. Alimoglu and E. Alpaydin, “Combining multiple representations and classifiers for pen-based handwritten digit recognition,” Proc. Int. Conf. Doc. Anal. Recognition, ICDAR, vol. 2, pp. 637–640, 1997, doi: 10.1109/ICDAR.1997.620583.

X. Liu, Y. Deng, Y. Sun, and Y. Zhou, “Multi-digit recognition with convolutional neural network and long short-term memory,” ICNC-FSKD 2018 - 14th Int. Conf. Nat. Comput. Fuzzy Syst. Knowl. Discov., pp. 1187–1192, Jul. 2018, doi: 10.1109/FSKD.2018.8686963.

S. Ahlawat and A. Choudhary, “Hybrid CNN-SVM Classifier for Handwritten Digit Recognition,” Procedia Comput. Sci., vol. 167, pp. 2554–2560, Jan. 2020, doi: 10.1016/J.PROCS.2020.03.309.

H. huang Zhao and H. Liu, “Multiple classifiers fusion and CNN feature extraction for handwritten digits recognition,” Granul. Comput., vol. 5, no. 3, pp. 411–418, Jul. 2020, doi: 10.1007/S41066-019-00158-6/TABLES/1.

J. Li, G. Sun, L. Yi, Q. Cao, F. Liang, and Y. Sun, “Handwritten Digit Recognition System Based on Convolutional Neural Network,” Proc. 2020 IEEE Int. Conf. Adv. Electr. Eng. Comput. Appl. AEECA 2020, pp. 739–742, Aug. 2020, doi: 10.1109/AEECA49918.2020.9213619.

R. Karakaya, S. Kazan, and S. Cakar, “Handwritten Digit Recognition Using Machine Learning,” Sak. Univ. J. Sci., vol. 25, no. 1, pp. 65–71, Feb. 2021, doi: 10.16984/SAUFENBILDER.801684.

M. Jain, G. Kaur, M. P. Quamar, and H. Gupta, “Handwritten digit recognition using CNN,” Proc. Int. Conf. Innov. Pract. Technol. Manag. ICIPTM 2021, pp. 211–215, Feb. 2021, doi: 10.1109/ICIPTM52218.2021.9388351.

S. Albahli, M. Nawaz, A. Javed, and A. Irtaza, “An improved faster-RCNN model for handwritten character recognition,” Arab. J. Sci. Eng., vol. 46, no. 9, pp. 8509–8523, Sep. 2021, doi: 10.1007/S13369-021-05471-4/METRICS.

A. Baldominos, Y. Saez, and P. Isasi, “A Survey of Handwritten Character Recognition with MNIST and EMNIST,” Appl. Sci. 2019, Vol. 9, Page 3169, vol. 9, no. 15, p. 3169, Aug. 2019, doi: 10.3390/APP9153169.

R. Raj, J. Mathew, S. K. Kannath, and J. Rajan, “Crossover based technique for data augmentation,” Comput. Methods Programs Biomed., vol. 218, p. 106716, May 2022, doi: 10.1016/J.CMPB.2022.106716.

A. Paszke et al., “PyTorch: An Imperative Style, High-Performance Deep Learning Library,” Adv. Neural Inf. Process. Syst., vol. 32, Dec. 2019, Accessed: May 25, 2024. [Online]. Available: https://arxiv.org/abs/1912.01703v1

“MNIST handwritten digit database, Yann LeCun, Corinna Cortes and Chris Burges.” Accessed: May 25, 2024. [Online]. Available: http://yann.lecun.com/exdb/mnist/.