Handwritten Hindu Numrals Recogonition using Support Vector Machine


By Ashwag M. Gadeed, Dr. Mohamed E. M. Musa.


Automatic Reading system is the technology that will enable us to get rid from Keyboards and allows the computer reads our writing. This paper introduces a new large data set for Arabic numeral recognition. The paper also describes the performance of a One Against All (OAA) Support Vector Machine (SVM)trained on this new data set. While the result of the classification is not very good, the experiments show the main problems of Hindu digit classification that such classifier could face. The analysis of the results reflects that the accuracy of this classifier could easily be enhanced to obtain much better result. As the data set is very new future work could show much better results using this data set.