Enhancing the Power of CNN Using Data Augmentation Techniques for Odia Handwritten Character Recognition

The performance of any machine learning model largely depends on the type input data provided. higher volume and variety data, better models get trained, thereby producing more accurate results. However, it is a challenging task to high in some cases containing enough variety. Handwritten character recognition for Odia language one them. NITROHCS v1.0 handwritten characters ISI image database numerals are standard datasets available research community. This paper shows five different that uses convolutional neural network identify response manipulated expanded using several augmentation techniques create variation increase given dataset. These models, with discussed paper, even lead further accuracy by approximately 1% across models. claims supported results from experiments done proposed numeral set.