Academic Certificate Verifier with Machine Learning

Abstract

It is getting scary how easy it is to fake a diploma now, and honestly, it just devalues all the work real students put in. We decided to do something about it by building a tool that uses AI to basically \"eye-test\" certificates for things like dodgy watermarks or fake signatures. Once a degree passes the test, we upload that data to a blockchain. Since that\'s permanent, the records stay tamper-proof forever. We even built a web portal so employers can verify a whole pile of documents in one go instead of doing it by hand. Our prototype has been working great so far. It catches the fakes without giving regular graduates a hard time, making the whole process way more secure for everyone.

Introduction

Academic certificates are widely used for education and employment verification, but their digital transformation has also made them vulnerable to forgery due to advanced editing tools. Fake certificates can now be created with high precision, making manual verification unreliable. To address this, technologies such as Optical Character Recognition (OCR), machine learning, deep learning, and blockchain are being explored to improve authenticity checks. OCR extracts text from certificates, AI models detect visual tampering, and blockchain ensures secure, tamper-proof storage of verified records.

Despite existing methods like OCR-based verification and QR code validation, there is still no unified system capable of detecting both textual and visual forgeries effectively, especially in cases involving altered signatures, seals, or grades. The proposed “Academic Certificate Authenticator” aims to fill this gap by combining OCR, deep learning-based forgery detection, and blockchain into a single integrated system. This improves accuracy, scalability, and reliability while reducing manual effort.

The system follows a modular pipeline where certificates undergo preprocessing, OCR extraction, AI-based visual analysis, and final validation. The dataset includes real, fake, and synthetically generated certificates, with preprocessing steps applied to enhance image quality and standardize inputs. Deep learning models then analyze fine visual details such as seals, signatures, and logos to detect tampering and assign a forgery probability score.

Conclusion

To wrap things up, we’ve built a solid system that tackles the massive problem of fake degrees head-on. By bringing together tools like smart scanning, AI \"forensics,\" and blockchain, we’ve created a way to verify certificates that is actually reliable and doesn\'t require a person to check every single detail by hand. Our approach is a huge improvement over the old-school ways of doing things. It\'s faster, more accurate, and can handle a huge volume of checks without breaking a sweat. Because we look at both the text and the visual \"red flags\"—like dodgy logos or seals—we can catch forgeries that used to slip through the cracks. Plus, using blockchain means the records are locked tight and can\'t be messed with, which builds a ton of trust in the results. Of course, there are still some hurdles, like needing good-quality scans and getting more schools to sync their data with our blockchain. But this is just the beginning. Looking ahead, we\'re planning to make the AI even sharper by feeding it way more data, and we\'re even thinking about launching a mobile app so people can run checks right from their phones. At the end of the day, this whole project is really about making sure that someone\'s hard-earned degree actually stands for something and keeps the entire system honest.

References

[1] R. B. Fisher, A. Yilmaz, and J. Kittler, “Document forgery detection using multimodal image analysis,” in Proc. Int. Conf. Document Analysis and Recognition (ICDAR), Lausanne, Switzerland, Sep. 2021, pp. 122–129, doi: 10.1109/ICDAR52620.2021.00125. [2] L. Baroffio, G. Valenzise, and M. Tagliasacchi, “Deep learning for signature and stamp verification in official documents,” in Proc. IEEE Int. Conf. Image Processing (ICIP), Bordeaux, France, Oct. 2022, pp. 2874–2878, doi: 10.1109/ICIP46576.2022.9897402. [3] C. Grech and A. Camilleri, “Blockcerts: A blockchain framework for secure and transparent academic credential verification,” IEEE Access, vol. 9, pp. 15632–15644, Mar. 2021, doi: 10.1109/ACCESS.2021.3059876. [4] T. Nguyen, P. Le, and D. Tran, “Digital watermarking techniques for certificate validation: A deep learning approach,” in Proc. IEEE Int. Conf. Information Security (ICIS), Seoul, South Korea, Nov. 2022, pp. 450–455, doi: 10.1109/ICIS56722.2022.10033451. [5] M. Mezzanotte, F. Magni, and A. Neri, “QR code–based systems for verifiable student academic records,” in Proc. IEEE Global Engineering Education Conf. (EDUCON), Vienna, Austria, Apr. 2021, pp. 225–230, doi: 10.1109/EDUCON46332.2021.9453998. [6] H. Schaefer and L. Müller, “Hybrid OCR and convolutional networks for robust document forgery detection,” IEEE Trans. Information Forensics and Security, vol. 17, pp. 3002–3015, Oct. 2022, doi: 10.1109/TIFS.2022.3200105. [7] A. Ojo, M. Adebayo, and K. Okonkwo, “Blockchain-based academic certificate verification in developing nations,” in Proc. IEEE Int. Conf. Blockchain, Sydney, Australia, Dec. 2021, pp. 201–208, doi: 10.1109/Blockchain53845.2021.9654399. [8] J. Vermaelen, S. Rossi, and G. Costa, “AI-driven anomaly detection for academic certificate authentication,” in Proc. IEEE Int. Conf. Machine Learning and Applications (ICMLA), Nassau, Bahamas, Dec. 2023, pp. 678–684, doi: 10.1109/ICMLA56720.2023.10124712. [9] R. Smith, “An overview of the Tesseract OCR engine,” in Proc. Int. Conf. Document Analysis and Recognition (ICDAR), Curitiba, Brazil, Sep. 2007, pp. 629–633, doi: 10.1109/ICDAR.2007.4376991. [10] D. Gruner, M. Power, and H. Zhang, “Blockchain-based digital credentialing systems: A survey and future directions,” IEEE Access, vol. 10, pp. 84532–84550, Aug. 2022, doi: 10.1109/ACCESS.2022.3198765.

Copyright

Copyright © 2026 Abhishek Kumar, Anand Pandey, Abhinav Srivastav, Abhishek Singh, Gaganjot Kaur. 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.