Survey on Machine Learning and Deep Learning Algorithms for Recognition of Handwritten Devanagari Characters

The automation of an application is crucial in the modern digital environment. In the atomization process, handwritten papers must be converted into digital format. There are more than 120 spoken languages that employ the Devanagari script. Recognizing handwritten Devanagari characters is made more challenging by the characters' complicated shapes, a wide range of modifiers, and variable writing styles. Preprocessing is a crucial step in recognizing handwritten images and comprises noise reduction, segmentation, feature extraction from the image, and more. Preprocessing and recognition are done using several methods. Due to its automatic feature extraction, a deep learning algorithm like CNN is particularly effective in recognition tasks. The main goal of this paper is to conduct a study of the literature and conduct a comparative analysis of the various preprocessing and recognition strategies to help guide future research on the problem of handwritten Devanagari script recognition