FAYEK Maarfi
فايق معارفي
fayek.maarfi@univ-msila.dz
0771644325
- Departement of ELECTRONICS
- Faculty of Technology
- Grade PHd
About Me
Science et Technologies
Research Domains
PHD RESEAUX ET TELECOM
FiliereTélécommunications
Location
Msila, Msila
Msila, ALGERIA
Code RFIDE- 1978-09-13 00:00:00
-
FAYEK Maarfi birthday
- 2025-11-25
-
2025-11-25
The 2nd International Workshop on Machine Learning and Deep Learning (WMLDL25)
Driver fatigue remains one of the most critical factors in preventable road deaths, yet conventional systems based on Convolutional Neural Networks (CNNs) often struggle to strike a vital balance between accuracy, speed, and practical usability under diverse conditions. This paper introduces WakeUp AI, an intelligent fatigue detection system explicitly designed to bridge that gap. The core framework leverages the advanced feature extraction capabilities of a Vision Transformer (ViT), combined with an optimized Support Vector Machine (SVM) classifier, resulting in an outstanding 99.82% test accuracy on the CEW dataset. This hybrid ViT-SVM approach achieves superior feature discrimination while maintaining computational efficiency suitable for edge deployment. For real-time use, WakeUp AI integrates MediaPipe FaceMesh with a streamlined ViT model, achieving inference latency of < 100ms frame. Crucially, a continuous temporal logic module constantly monitors the driver’s eye state, activating instant audio alerts only when fatigue patterns (such as prolonged eye closure duration) are robustly detected. Unlike conventional systems limited to simple, rigid thresholding, WakeUp AI intelligently adapts to diverse environments, making it exceptionally robust. By combining state-of-the-art deep learning with real-time responsiveness, WakeUp AI offers a scalable, high-performance solution for critical safety applications.
Citation
FAYEK Maarfi , BRIK Youcef , BILAL Attallah , ,(2025-11-25), The 2nd International Workshop on Machine Learning and Deep Learning (WMLDL25),WakeUp AI – Fatigue Detection System,Msila, Algeria
- 2025-11-25
-
2025-11-25
A Hybrid ResNet50-PCA-SVM Model for Accurate Palmprint Recognition
Biometric authentication has demonstrated effectiveness in accurately verifying an individual’s identity. Specifically, palmprint-based biometric systems have gained growing interest in recent years because of their high security, feasibility, and user acceptance. Conventional palmprint recognition approaches involve extracting palmprint features prior to classification, which can influence the recognition performance. In this study, we propose a hybrid model combining deep learning and traditional machine learning techniques for palmprint identification. Features are first extracted using a pre-trained ResNet50 model, then reduced via Principal Component Analysis (PCA), and finally classified using a Support Vector Machine (SVM) with a linear kernel. The experiments are conducted on publicly available contactless datasets, namely Tongji and BMPD, under different data split ratios. Performance is evaluated using accuracy, precision, recall, F1-score, and ROC-AUC analysis. The experimental results show that the proposed approach achieves an accuracy exceeding 99.3% across all tests, with an AUC of 1.00, demonstrating the effectiveness of combining CNN-based feature extraction with classical machine learning techniques for palmprint recognition.
Citation
KHEIRA FAIROUZ Bedjekina , BILAL Attallah , BRIK Youcef , FAYEK Maarfi , ,(2025-11-25), A Hybrid ResNet50-PCA-SVM Model for Accurate Palmprint Recognition,The 2nd International Workshop on Machine Learning and Deep Learning (WMLDL25),University of Mohamed Boudiaf of M'sila
- 2024-07-19
-
2024-07-19
Facial Recognition: Employing Histogram of Gradients and Support Vector Machines
Abstract – Facial recognition systems are integral to modern applications such as security, access control, and user authentication. This study explores the implementation of a facial recognition system employing Histogram of Gradients (HOG) and Support Vector Machines (SVM). HOG features are effective in capturing fine details of facial structures, while SVMs provide robust classification capabilities. The selection of appropriate features and classifiers significantly impacts system performance. In this work, we propose and evaluate a system using HOG features and SVMs, leveraging their synergy for accurate and reliable face recognition. Experimental results demonstrate the effectiveness of this approach, showcasing its potential for enhancing security and user interaction in various domains. Keywords – Face Recognition, Support Vector Machine (SVM),Histogram Of Gradients(HOG), Machine Learning, Classification, Feature extraction
Citation
FAYEK Maarfi , ,(2024-07-19), Facial Recognition: Employing Histogram of Gradients and Support Vector Machines,4th International Conference on Scientific and Academic Research ICSAR 2024,Konya, Turkey
- 2024-07-19
-
2024-07-19
Integrating Haar Cascade, CNN and SVM for Advanced Facial Recognition
Abstract – Facial recognition systems have become integral in various applications, ranging from security and access control to user authentication. This study presents a robust facial recognition system leveraging the combined strengths of Haar Cascade, Convolutional Neural Networks (CNN), and Support Vector Machines (SVM). The system begins with face detection using Haar Cascade, a well-established technique for identifying facial regions in images. Following detection, the detected faces are processed through a CNN, specifically the VGG16 model, for feature extraction. The high-level features extracted by the CNN are then classified using an SVM, which excels in handling high-dimensional data and providing precise classification. The proposed system undergoes rigorous training and evaluation using a comprehensive dataset of facial images organized by individual directories. The dataset is split into training and testing sets to validate the model's performance. Experimental results demonstrate the efficacy of this approach, with the SVM classifier achieving high accuracy in facial recognition tasks. In addition to evaluating the system's performance, the study includes a visual inspection of the results through plots of test images, showcasing actual and predicted labels. This highlights the model's ability to accurately identify and verify individuals based on facial features. The integration of Haar Cascade, CNN, and SVM in this facial recognition system underscores the importance of combining different methodologies to enhance accuracy and reliability. This study contributes to the development of advanced facial recognition technologies, providing a valuable tool for applications requiring precise identification and verification. Keywords – Face Recognition, Support Vector Machine (SVM), Haar Cascade, Convolutional Neural Network (CNN), Feature extraction
Citation
FAYEK Maarfi , ,(2024-07-19), Integrating Haar Cascade, CNN and SVM for Advanced Facial Recognition,4th International Conference on Scientific and Academic Research ICSAR 2024,Konya, Turkey