Email Security using Facial Recognition

Abstract

The project aims to bolster email security by employing advanced facial recognition technology. Leveraging OpenCV and convolutional neural networks (CNNs), the system verifies the identity of users attempting to access email accounts by analyzing facial biometrics. This approach enhances authentication measures, mitigates the risks of unauthorized access, and strengthens overall email privacy and confidentiality.Through real-time facial recognition algorithms, the system ensures secure access to email accounts, providing a robust defense against potential security breaches.

Introduction

The project enhances email security by using face recognition technology based on OpenCV, TensorFlow, and Convolutional Neural Networks (CNNs). It captures facial images of users, extracts features, and trains a deep learning model to accurately identify authorized individuals. When accessing email, the system matches the detected face against the trained database and grants access only if the confidence score meets a set threshold, otherwise denying access.

The development uses OpenCV for real-time image capture and preprocessing, while CNNs handle feature extraction and classification. Early facial recognition algorithms rely on Haar features, integral images, Adaboost for feature selection, and cascade classifiers to speed up detection.

The system provides a biometric-based security layer that improves over time with more data, adapting to variations in lighting and expressions. This reduces risks related to password theft and unauthorized email access, enhancing privacy and cybersecurity.

The methodology includes hardware specs (Intel Core i3, 8GB RAM), software (Python, OpenCV, TensorFlow), and the CNN architecture with layers for convolution, activation (ReLU), pooling, and fully connected classification. Training involves capturing face images, building a dataset, matching faces, and alerting admins on unauthorized access, ensuring only verified users can access emails.

Conclusion

In this project, I extend the concept of Multiview Face Detection using Convolution Neural Networks to provide an automatic tagging system. Once all the faces are detected, they are tagged using Local Binary Patterns Histograms (LBPH) method. Furthermore, Precision, Recall and F-measure calculations show an accuracy of 85% for tagging the faces which are successfully detected. The live face detection and prediction can be implemented in a successive manner by using OpenCV and Numpy Library. This implementation helps the users to maintain the locker in safe and secure manner. Any unauthorized person access will be intimated to the account user by sending a mail. The proposed work and model creation also be extended to use it with any CCTV footage system for identifying people in case of theft, murder, etc. More specifically, it could even be used by the law enforcement agencies to identify absconding criminals in case they are spotted at the airport, railway station or bus-station by these surveillance cameras.

References

[1] ”Automate The Boring Stuff With Python, 2Nd Edition: Practical Programming For Total Beginner”, SL Swegart, 2020. [2] “Learning Python 5ed: Powerful Object-Oriented Programming” O’Relly – 2014. [3] “Elements of Programming Interviews in Python: The Insiders\' Guide”, Adan Aziz, Tsung-Hsen Lee – 2021.

Copyright

Copyright © 2025 Gangalakshmi P, Swetha S, Sharmila B, Mrs. G. Sumathi. 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.