This paper introduces a face recognition system designed to tackle the issue of occlusion, which can hinder the effectiveness of traditional facial recognition systems when parts of a face are obstructed. To address this, the system is built using object-oriented analysis and Design Methodology (OODAM), which promotes a modular and flexible development approach. The implementation is carried out in Python, utilizing the Deep Face library for face recognition.in particular, the system employs the “Facenet512” Model with the “Euclidean_12” distance metric to enhance accuracy in face identification, even in the presence of partial occlusions. Machine learning algorithms are integrated for feature extraction and matching, with a SQLite database used for storing and managing face efficiently. The architecture supports real-time detection and recognition of faces through OpenCV, while a kernelized correlation filters (KCF) tracker is used to ensure continuous tracking across video frames. The system also processes frames, addresses occlusions, and stores recognition outcomes in a well-organized database. Evaluation result highlights the system’s capacity to effectively mitigate the impact of occlusions, achieving improved recognition accuracy and reliability over traditional methods.
Machine Learning, Face Recognition, Occlusion, Real-time Detection
IRE Journals:
Chendo Chukwunonso Nnamdi , Okeke Ogochukwu , Mgbeafulike Ike , Okafor Patrick
"Real-Time Face Recognition with Occlusion Handling Using Facenet512 and Machine Learning" Iconic Research And Engineering Journals Volume 8 Issue 10 2025 Page 526-534
IEEE:
Chendo Chukwunonso Nnamdi , Okeke Ogochukwu , Mgbeafulike Ike , Okafor Patrick
"Real-Time Face Recognition with Occlusion Handling Using Facenet512 and Machine Learning" Iconic Research And Engineering Journals, 8(10)