Comparison of deep CNN and ResNet for Handwritten Devanagari Character Recognition

Abstract

Handwritten Optical Character Recognition is a lush area of research and is used in various real time applications. This research is based on comparative analysis of handwritten OCR by using Deep CNN and ResNet for Devanagari script, a regional language. Devanagari character contains two elements, diacritics and the main grapheme. Key challenge associated with Devanagari script is many a time different characters look similar. Secondly some characters are written differently by different individuals. Proposed ResNet manages vanishing gradient issue and improves capability of traditional Deep CNN. It uses dynamic flow of activation. ResNet identity blocks help in to overcome vanishing gradient issues. Proposed architecture scored close to 99% accuracy for the DHCD, which is better than other state-of-art results. Training phase of proposed model is reasonably less than many other variants of deep CNN.