AI-Powered Automated Evaluation of Handwritten Student Answers Using OCR and BERT-Based Semantic Similarity

Marking of handwritten answer sheet is also time consuming, inconsistent and subject to human bias. The AI Answer Sheet Evaluator solves these issues by presenting a new model of automated grading with Artificial Intelligence (AI) and Natural Language Processing (NLP). The system uses OCR programs like Tesseract and TrOCR to turn the handwritten student answers into a digital text. Semantic similarity models such as Sentence-BERT and GPT are then used to evaluate extracted text by comparing answers of students with answers given by the instructor using the models. Judging is done according to accuracy, relevancy, completeness and coverage of important points. This model is unlike traditional systems that only provide binary scoring since it supports partial marking whereby fair distribution of mark can be made to partly correct answers. With OCR and NLP, the system is effective in handling big amounts of scripts in addition to being transparent and objective. This smart and scalable solution will help bring about a transformation in the field of academic assessment in schools, universities, and extensive test-taking facilities.