AI-Powered Handwritten Exam Evaluation System Using OCR and Text Similarity

ABSTRACT In an era where digital transformation is reshaping traditional educational practices, the need for automation in examination assessment has become increasingly vital. This research introduces an innovative AI-powered system designed to automate the evaluation of handwritten exam papers by integrating Optical Character Recognition (OCR) and Natural Language Processing (NLP) techniques. The system uses advanced OCR models like TrOCR and EasyOCR to extract content from scanned answer scripts and then employs text similarity techniques, such as cosine similarity and transformer-based models, to compare the extracted answers with a reference key. The final result displays the extracted response, similarity percentage, and the marks assigned. This approach enhances accuracy, reduces evaluator workload, and promotes transparency in assessment. Keywords – OCR, NLP, TrOCR, EasyOCR, Cosine Similarity, Text Evaluation, AI in Education, Handwritten Answer Evaluation.