Determining the Question Based on Handwriting-Recognition of Answer Scripts

Abstract

The most used form of testing even in modern days is pen-and-paper. In a world where digitizing is rapidly overtaking all other means, it is imperative that we modernize our examination techniques. Answers are split based on question numbers and these are corrected by evaluators. In case of errors in splitting due to incorrect marking or scanning, handwriting recognition can be used to recognize the question an answer belongs to. This approach has been dwelled into in this paper. By creating a personalized dataset for each batch of students, this approach seems promising and gives good results.