NOVEL FACIAL RECOGNITION AND ATTENDANCE LOGGING: A ROBUST PIPELINE COMBINING DEEP LEARNING AND META-LEARNER

The ineffectiveness of conventional approaches has led to a considerable increase in the need for automated attendance management systems in business and educational settings. For precise and effective attendance tracking, this study offers a real-time facial recognitionbased system that makes use of transfer learning strategies. For face feature extraction, the system makes use of pre-trained ResNet34 and EfficientNet models, which are then improved via a meta-learner module to improve predictions. Real-time photos are captured using a live webcam interface, which automatically registers attendance by comparing them to a prestored dataset. The system also keeps track of entry and exit timestamps, which makes it possible to determine how long each student stays in class. This technology dramatically increases accuracy, scalability, and real-time performance when compared to traditional techniques like Haar Cascade and KNN-based face recognition. The incorporation of transfer learning guarantees flexibility in a variety of scenarios, rendering the system a resilient and expandable attendance solution.