Neural Network with Regression Algorithms for Optical Character Recognition


  • J V.Vidhya
  • Archit Kumar
  • Yashwardhan Parakh





Optical Character Recognition, deep Neural network, TensorFlow, patten matching, machine learning, regression, gradient descent, softmax regression


In today's automatic and robust modern world, possibilities of optical character recognition is endless. Previously OCR was used in postal service to read address from mail, car number plate tracking, automation of bank check transfer but today it has taken document management system to whole new level. Using OCR we can convert normal hardcopy document into Searchable text. We will use deep Neural network to train systems to recognise characters in a precise manner, basically we have proposed neural network model combined with machine learning technique like gradientDescent, regression, softmax normalization which will help to increase the efficiency of the OCR. Computer will able to recognise hand written digit. We will be using Google's advanced TensorFlow to create an OCR system which will be efficient and robust in action.




[1] A Neural Network Approach to Character Recognition (Rajavelu, Musavi and Shirvaikar).

[2] Spatially sparse Convolutional neural network (Benjamin Graham).







View Full Article:

How to Cite

V.Vidhya, J., Kumar, A., & Parakh, Y. (2018). Neural Network with Regression Algorithms for Optical Character Recognition. International Journal of Engineering & Technology, 7(3.12), 341–343.
Received 2018-07-23
Accepted 2018-07-23
Published 2018-07-20