Abstract
Character recognition is a very popular machine learning application for example—OMR sheets are commonly used for grading purposes all over the world. It is a procedure of analysing human-marked data from documents like surveys and tests. The COVID-19 epidemic has had a negative influence on schooling, since it has caused students’ handwriting habits to deteriorate significantly. As a result, the goal of this paper is to propose a lightweight deep learning model that uses computer vision to aid in the practice, evaluation, and analysis of nursery students’ handwriting by evaluating sheets and pointing out major errors with an accuracy score that can be used to measure improvement. Recognition, evaluation, and analysis are the three primary components of the suggested paradigm. The steps are as follows: (1) Pre-processing, (2) deep learning model for feature extraction and classification, (3) predicting the letters in the supplied picture, (4) calculating the similarity index between the input image and the glyphs in the computer font, (5) matching the input image's stroke length and angles to font glyphs, and (6) displaying the character's score and analysis. The most accurate model for this process was 83.04%.