ONE SOLUTION OF OPTICAL CHARACTER RECOGNITION USING DEEP NEURAL NETWORKS

Authors

  • Dejan Ikonić Autor

DOI:

https://doi.org/10.24867/16BE31Ikonic

Keywords:

Deep learning, Artificial intelligence, Neural networks

Abstract

This paper presents one approach of automated optical character recognition on photographs with streetlights identification using deep Convolutional Neural Network. The task was realized in three iterations, with an accuracy of 93.04%.

References

[1] C. Tanprasert, T. Koanantakool: "Thai OCR: a neural network application", Proceedings of Digital Processing Applications, vol. 1, pp. 90-95, Perth, Australia, 1996.
[2] S. Ali, Z. Shaukat, M. Azeem, Z. Sakhawat, T. Mahmood, K. ur Rehman: "An efficient and improved scheme for handwritten digit recognition based on convolutional neural network", SN Applied Sciences, Springer Nature Switzerland, vol. 1, art. 1125, pp. 1-9, 2019.
[3] D.C. Ciresan, A. Giusti, L.M. Gambardella, J. Schmidhuber: "Deep Neural Networks Segment Neuronal Membranes in Electron Microscopy Images", Advances in neural information processing systems, Lake Tahoe, Nevada, United States, 2012.
[4] S. Gavran: "Veštačke neuronske mreže u istraživanju podataka: pregled i primena", Univerzitet u Beogradu, Matematički fakultet, Master rad, 2016.
[5] https://github.com/mdbloice/Augmentor (pristupljeno u septembru 2021.)
[6] A. Krizhevsky, I. Sutskever, G.E. Hinton: "Imagenet classification with deep convolutional neural networks", Communications of the ACM, Association for Computing Machinery, vol. 60, issue 6, pp. 84–90, 2017.
[7] S. Hassanpour, N. Tomita, T. DeLise, B. Crosier, L. A. Marsch: "Identifying substance use risk based on deep neural networks and Instagram social media data", Neuropsychopharmacol, Springer Nature, vol. 44, pp. 487–494, 2019.

Published

2022-02-04

Issue

Section

Electrotechnical and Computer Engineering