Abstract: Recognition of handwritten digits has been an important area in recent years because of its uses in many fields. Arabic pattern digits, weak work is performed because Arabic digits (Indian) are more complicated than English patterns. This study focuses on the recognition component of the recognition of handwritten Arabic digits (Indian) that faces many obstacles, including the infinite variety of human handwriting and the broad public databases. The study presented a deep learning approach that can effectively be applied to the recognition of handwritten Arabic digits. Convolutional Neural Network (CNN) trained and tested MADBase database (Arabic handwritten digits images) with 60000 training and 10000 test images.. A contrast is made between the results, and it is seen at the end that the use of CNN has resulted in substantial improvements across various classification algorithms for machine learning, As a test accuracy with better results than other approaches using the same database, the test accuracy was improved to 99.25%.
Keywords: Recognition, handwritten, classification, Convolutional Neural Network.
|
DOI:
10.17148/IJARCCE.2020.91109
