CNRR- Net: A Hybrid Deep Learning Framework for Fake License Plate Detection Using YOLOv8, OCR, and Vehicle Identity Verification

Abstract

The increasing frequency of forged, altered, and duplicated vehicle license plates has resulted in serious chal-lenges for intelligent transportation systems and law-enforcement agencies. Traditional approaches to the verification of vehicle registration details are inefficient and not suitable for real-time applications, while traditional ANPR systems mainly focus on the recognition of a number plate and fail to identify the mismatch between plate information and the actual vehicle. This work proposes an automated fake number plate detection system that integrates deep learning-based license plate detection, OCR-driven text extraction, and database-level verification. A YOLO-based detection module has been employed for accurate localization of the number plates in an image of diverse environmental conditions; this is followed by robust pre-processing and OCR, which transforms the plate characters into machine-readable text. The extracted registration number is then cross-checked with a secure back-end database containing the chassis number, model name, and year of manufacture to look for inconsistencies that may indicate tampering or fraudulent usage. Experimental evaluations establish that the proposed approach achieves high accuracy in recognition and authenticity verification and thus effectively differentiates between genuine plates from fake/mismatched ones. The proposed system is scalable and can be deployed in real time for intelligent traffic monitoring, improved road safety, and law-enforcement applications.

Introduction

The text discusses the growing problem of fake or altered vehicle license plates in Intelligent Transportation Systems (ITS) and the need for more advanced detection methods. Traditional manual verification and basic ANPR systems are insufficient because they mainly focus on plate recognition but do not verify authenticity, making them vulnerable to fraud, misuse, and evasion of law enforcement.

To address this gap, the paper proposes a deep learning–based system that combines YOLO-based object detection, OCR (Optical Character Recognition), image preprocessing, and database verification. YOLO models are used to detect license plates and vehicle features in real time, while OCR extracts the text from plates. The extracted data is then cross-checked with a secure vehicle database containing details like chassis number, model, and manufacturing year to verify authenticity.

The methodology includes several steps: license plate detection, image preprocessing to improve OCR accuracy, character recognition using OCR tools like EasyOCR and Tesseract, database verification, and final fraud classification (genuine, fake, or suspicious). The system also uses vehicle attribute matching to detect inconsistencies, such as mismatches between plate data and vehicle type.

The literature review shows that most existing research focuses on improving detection and recognition accuracy but largely ignores fake plate identification. This highlights a research gap that the proposed system aims to fill.

Results show that the YOLOv8-based system achieves high accuracy (about 97.2% mAP) with real-time performance, while OCR accuracy improves significantly with preprocessing techniques. Overall, the system demonstrates strong potential for improving road safety, traffic monitoring, and law enforcement by detecting fraudulent number plates more effectively than traditional ANPR systems.

Conclusion

The proposed Fake Number Plate Detection System is the brainchild of senior capstone project team. This innovation depicts the utilization of tech, namely vision and verification by database, to figure out phony or nonmatching vehicle license plates. It is leveraging YOLOv8 for localization of the plate, EasyOCR for recognition of the characters with struc-tured database validation based on the chassis number, model, and manufacturing details, the system reaches an accuracy of 96.4% as indicated in Table II. The hybrid method used here to lessen the occurrence of false positives due to plate duplication or tampering thus ensures that the reliability level is higher than the one obtained by the vision-only systems. From the same source of truth, it is evident that the experimental findings confirm that preprocessing and adaptive OCR are two factors that can greatly improve recognition resilience in the scenarios that occur in the real world, and which include dim lighting, motion blur, and occlusion. Since the recognized license plates are legitimate, then the cross-checking with the database is what makes the system even stronger. Next steps will focus on the inclusion of cloud-based APIs for vehicle verification, expanding the dataset for various local formats, and real-time deployment using edge devices. Therefore, as a tool for automated car monitoring and law enforcement, this solution is viable, scalable, and secure.

References

[1] N. Bharti, M. Kumar and V. M. Manikandan, ”A Hybrid System with Number Plate Recognition and Vehicle Type Identification for Vehicle Authentication at the Restricted Premises,” 2022 2nd Interna-tional Conference on Emerging Frontiers in Electrical and Electronic Technologies (ICEFEET), Patna, India, 2022, pp. 1-6, doi: 10.1109/ICE-FEET51821.2022.9847975. [2] L. Khedekar, S. Kadu, C. Kakde, P. Kalawadiya, J. Khobragade and A Kalaskar, ”Automated Number Plate Analysis: Using Yolov8 and Easyocr for Enchanted Surveillance,” 2025 9th International Conference on Inventive Systems and Control (ICISC), Coimbatore, India, 2025, pp. 1-7, doi: 10.1109/ICISC65841.2025.11188381. [3] N. B. Bahadure et al., ”Computer vision-based real-time vehicle and traf-fic management and classification system,” 2025 4th OPJU International Technology Conference (OTCON) on Smart Computing for Innovation and Advancement in Industry 5.0, Raigarh, India, 2025, pp. 1-6, doi: 10.1109/OTCON65728.2025.11071088. [4] N. Shelke, S. Jadhav, M. Doifode, Y. Umate, R. Patil and N. Harinkhede, ”A Review on Identification of Number Plate and Wrong Way Vehicles Detection,” 2023 IEEE International Students’ Conference on Electrical, Electronics and Computer Science (SCEECS), Bhopal, India, 2023, pp. 1-4, doi: 10.1109/SCEECS57921.2023.10063054. [5] S. R, D. S, H. T, D. J. E and A. S, ”YOLO-based Automatic Number Plate Recognition with Unauthorized Font Detection,” 2025 6th Inter-national Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV), Tirunelveli, India, 2025, pp. 722-727, doi: 10.1109/ICICV64824.2025.11085826. [6] Z. F. Al Fitroh and G. Ariyanto, ”Implementation of YOLOv11 for Vehicle License Plate Detection and Recognition,” 2025 International Conference on Smart Computing, IoT and Machine Learning (SIML), Surakarta, Indonesia, 2025, pp. 1-6, doi: 10.1109/SIML65326.2025.11081138. [7] N. Anand, S. Sunny, G. Priya, G. Krishna and P. Jyothi, ”Auto-matic Number Plate Recognition Powered by Deep Learning,” 2025 2nd International Conference on Trends in Engineering Systems and Technologies (ICTEST), Ernakulam, India, 2025, pp. 1-6, doi: 10.1109/ICTEST64710.2025.11042683. [8] I. Malik, S. Verma, S. Sachan and Rohit, ”YOLOv7-Based Hel-met Detection System For Traffic Safety,” 2024 Sixth Interna-tional Conference on Computational Intelligence and Communica-tion Technologies (CCICT), Sonepat, India, 2024, pp. 493-497, doi: 10.1109/CCICT62777.2024.00082. [9] R. Garg, R. Wason, P. Arora and M. N. Hoda, ”Robust Advanced Number Plate Recognition (ANPR) System from Real-Time Video Footage,” 2025 12th International Conference on Computing for Sus-tainable Global Development (INDIACom), Delhi, India, 2025, pp. 1-4, doi: 10.23919/INDIACom66777.2025.11115864. [10] A. K. Sandhu, ”A Novel Framework for Image Denoising for Mixed Noise Using Bilateral and NSST,” 2025 12th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions) (ICRITO), Noida NCR, India, 2025, pp. 1-5, doi: 10.1109/ICRITO66076.2025.11241249. [11] R. Shashidhar, A. S. Manjunath, R. Santhosh Kumar, M. Roopa and S. B. Puneeth, ”Vehicle Number Plate Detection and Recognition using YOLO- V3 and OCR Method,” 2021 IEEE International Con-ference on Mobile Networks and Wireless Communications (ICM-NWC), Tumkur, Karnataka, India, 2021, pp. 1-5, doi: 10.1109/ICM-NWC52512.2021.9688407. [12] Qiang Huang, Kejie Li and Y. Nakamura, ”Humanoid walk control with feedforward dynamic pattern and feedback sensory reflection,” Proceedings 2001 IEEE International Symposium on Computational Intelligence in Robotics and Automation (Cat. No.01EX515), Banff, AB, Canada, 2001, pp. 29-34, doi: 10.1109/CIRA.2001.1013168. [13] T. Mustafa and M. Karabatak, ”Challenges in Automatic License Plate Recognition System Review,” 2023 11th International Symposium on Digital Forensics and Security (ISDFS), Chattanooga, TN, USA, 2023, [14] pp. 1-6, doi: 10.1109/ISDFS58141.2023.10131688. [15] M. A. Riyadi, M. S. Tawa’qi and I. Setiawan, ”Design of Vehicle License Plate Detection System Using YOLOv8 and PaddleOCR,” 2025 4th International Conference on Electronics Representation and Algorithm (ICERA), Yogyakarta, Indonesia, 2025, pp. 98-102, doi: 10.1109/ICERA66156.2025.11087294. [16] H. K. Dash et al., ”Automatic Vehicle Number Plate Detection and Recognition System using YOLOv8n and EasyOCR,” 2024 International Conference on Vehicular Technology and Transporta-tion Systems (ICVTTS), Bangalore, India, 2024, pp. 1-6, doi: 10.1109/ICVTTS62812.2024.10763935. [17] V. V. R. SP, B. Sathyasri, A. Balaji, S. Vanaja, R. Krishnan and Y. Deepika, ”Automatic Number Plate Recognition System for Entry and Exit Management,” 2021 6th International Conference on Communica-tion and Electronics Systems (ICCES), Coimbatre, India, 2021, pp. 1-5, doi: 10.1109/ICCES51350.2021.9489018.

Copyright

Copyright © 2026 Smriti Nayak, Sonali Ganiga, Manvith Poojary. 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.