Hyper Parameter Optimization of CRNN for Printed Devanagari Script Recognition using Taguchi's Method

The Devanagari script is one of the most widely used scripts worldwide. The existing deep learning-based optical character recognition system for printed Devanagari scripts using Convolutional Neural Network - Recurrent Neural Network, or CRNN is not robust enough to recognize any randomly printed Devanagari scanned document. At present, the hyper-parameters of the CRNN system are selected randomly either with the trial-and-error or grid search methods. Moreover, there is no optimized way to choose the hyperparameters of the CRNN, which improves the recognition accuracy for Devanagari documents. Furthermore, the lack of standard Devanagari script datasets has hampered the development of word recognizers. In this paper, the hyper-parameter of the CRNN system is optimized using Taguchi's method of optimization. The performance of the hyper-parameters optimized CRNN system is compared with the current state-of-the-art text recognition CRNN network. The results reveal that the CRNN optimized with Taguchi's method performs better than the CRNN-based systems.

Automated summary

This paper focuses on the optimization problem of the optical character recognition system for printed Devanagari scripts. The current system faces challenges in recognizing scanned Devanagari documents, as well as issues with randomly selecting hyperparameters and a lack of standardized datasets. The study optimizes the hyperparameters using Taguchi's method and compares the performance with the state-of-the-art CRNN network, showing that the Taguchi optimized system outperforms CRNN-based systems.