Related Experiment Video
Updated: Jun 21, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
2.7K
Computationally intelligent real-time security surveillance system in the education sector using deep learning.
Muhammad Mobeen Abid1, Toqeer Mahmood1, Rahan Ashraf1
1Department of Computer Science, National Textile University, Faisalabad, Pakistan.
Plos One
|July 11, 2024
Summary
This study introduces an enhanced FaceNet model for accurate real-time facial recognition. Achieving 99.86% accuracy, it improves upon traditional methods for security surveillance.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Biometrics
Background:
- Traditional facial detection methods (Haar-like, MTCNN, AdaBoost) balance speed and accuracy using template matching and geometric features.
- Real-time security surveillance and identity matching necessitate advanced face detection and recognition techniques.
Purpose of the Study:
- To present an enhanced FaceNet network for high-accuracy, real-time face detection and recognition.
- To evaluate the proposed framework against existing traditional and deep learning techniques.
Main Methods:
- Employed RetinaFace for rapid face detection and alignment.
- Utilized an enhanced FaceNet with an improved loss function for accurate face verification and recognition.
- Conducted comparative evaluations of detection and recognition performance.
Main Results:
- The enhanced FaceNet framework meets real-time facial recognition requirements.
- Achieved a face recognition accuracy of 99.86%, fulfilling practical needs.
- Demonstrated superior performance compared to traditional and other deep learning methods.
Conclusions:
- The proposed enhanced FaceNet solution is effective for real-time security surveillance and identity matching.
- Shows significant potential for applications in the education sector requiring robust face detection and recognition.
