AI Driven Handwritten Assignment Analysis and Evaluation

Abstract

Manual evaluation of handwritten assignments is a time-consuming and labor-intensive process that burdens faculty members in educational institutions. This paper presents a novel AI-driven system designed to automate the evaluation of handwritten assignments using Optical Character Recognition (OCR) and Natural Language Processing (NLP). The system ensures accuracy, consistency, and efficiency in grading while reducing administrative workload. Unique assignment questions are generated from a robust question bank and distributed via institutional email. Students submit scanned handwritten responses through a dedicated portal, where OCR converts the images into editable text. An NLP engine then evaluates the submissions based on rubric-based grading criteria. Results are stored in an Excel sheet, readily accessible to faculty. This scalable solution enhances academic integrity, facilitates seamless evaluation, and can be integrated into modern educational workflows.

Introduction

This project proposes an AI-driven system to automate the evaluation of handwritten student assignments, addressing the faculty’s burden of manual grading and reducing errors. The system uses a dynamic question bank to generate unique assignments sent securely to students, who upload their handwritten responses via an online portal. Optical Character Recognition (OCR) converts the handwriting into text, which is then analyzed by Natural Language Processing (NLP) algorithms to evaluate answers against predefined rubrics, ensuring consistent and fair grading.

The system’s workflow includes submission, OCR processing, text cleaning, semantic comparison with model answers using techniques like BERT embeddings and cosine similarity, and automated scoring. Results are stored in a database and shared via dashboards and Excel sheets for faculty review. Faculty can manually adjust grades to improve the system’s learning over time.

Literature highlights advancements in NLP for essay grading and OCR for handwritten text recognition, which the system leverages to provide accurate, scalable, and context-aware assessment. The user interface offers a seamless experience for login, assignment upload, evaluation progress, and confirmation of successful submission.

Conclusion

The proposed AI-based system for automated evaluation of handwritten assignments effectively integrates Optical Character Recognition (OCR) and Natural Language Processing (NLP) to streamline the academic grading process. By leveraging OCR, the system accurately converts handwritten content into digital text, which is then analyzed using advanced NLP models to evaluate semantic similarity, grammatical structure, and content relevance. This approach not only reduces manual workload and human bias but also ensures faster, consistent, and scalable evaluation across a large volume of assignments. Furthermore, the integration of deep learning techniques enhances the system’s ability to handle diverse handwriting styles and complex answer patterns. Overall, this intelligent auto-grading system represents a significant step toward digital transformation in the education sector, offering a reliable and efficient solution for academic institutions aiming to modernize their assessment methodologies. The AI-based system is designed to be adaptable to various academic disciplines and subjects, enabling it to assess a wide range of assignments. By utilizing a predefined question bank, the system ensures that each student receives unique questions, promoting fairness and preventing cheating. The system\'s ability to dynamically adjust difficulty levels based on the student\'s responses further personalizes the assessment, providing a more accurate reflection of their knowledge and capabilities. Additionally, the system can continuously learn and improve its grading accuracy over time by incorporating feedback and fine-tuning the NLP models, enhancing its performance as more data is processed. The system\'s use of Natural Language Processing also ensures that the evaluation is context-aware and can detect rephrased answers, paraphrased content, or subtle differences in meaning, all of which are commonly encountered in academic assignments. It goes beyond simple keyword matching, enabling a deeper understanding of the student\'s thought process and reasoning. This approach not only results in more accurate and fair grading but also provides valuable insights into areas where students may need further improvement or clarification. As such, the system becomes a powerful tool for both grading and educational development. In addition to automating the grading process, the system also streamlines administrative tasks such as report generation and feedback distribution. Once the evaluation is complete, the results are automatically compiled into an easy-to-read format, which is then shared with faculty for review or directly with students. This eliminates the need for time-consuming manual entry of grades and feedback, reducing the administrative burden on educators and allowing them to focus more on teaching and student support.

References

[1] Sharma, R.; Mehta, A.; Gupta, S. \"AI-Based Automated Essay Grading System Using NLP.\" International Journal of Artificial Intelligence and Applications, vol. 9, no. 2, 2018, pp. 45–52. [2] Zhang, Y.; Lee, C.H.; Wong, K. \"Advancing Optical Character Recognition for Handwritten Text Using Deep Neural Networks.\" Pattern Recognition Letters, vol. 135, 2020, pp. 137–144. [3] Bansal, K.; Goyal, P. \"NLP-Based Automatic Answer Evaluation for Subjective Responses.\" International Journal of Advanced Computer Science and Applications, vol. 11, no. 4, 2020, pp. 201–209. [4] Verma, A.; Singh, M.; Kapoor, R. \"Improved Automated Essay Grading System via Natural Language Processing and Deep Learning.\" IEEE Access, vol. 8, 2020, pp. 155843–155854. [5] Narayan, S.; Patel, R.; Sinha, A. \"Intelligent Auto-Grading System for Descriptive Answers Using Machine Learning.\" Procedia Computer Science, vol. 172, 2020, pp. 1025–1032. [6] Rahman, A.; Qureshi, S.A.; Farooq, U. \"Optical Character Recognition (OCR) for Text Recognition and its Post Processing Method: A Literature Review.\" Journal of Computer Science & Systems Biology, vol. 15, no. 3, 2021, pp. 89–97. [7] Iyer, P.; Kulkarni, V.; Deshmukh, R. \"Enhanced Framework for Evaluating Handwritten Scripts Using Deep Learning.\" International Journal of Emerging Trends in Engineering Research, vol. 8, no. 7, 2020, pp. 3421–3428.

Copyright

Copyright © 2025 Sandeep Reddy Boreddy, Saketh Reddy Tadi, Eshwar Goud Mandala, G Udaya Sree. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.