Real-Time Face Attendance System Using CNN Mobilenet and MTCNN

This research presents the development of a real-time attendance system utilizing facial recognition, which incorporates three main components: the MobileNet Convolutional Neural Network (CNN) for classification, Multi-task Cascaded Convolutional Networks (MTCNN) for face detection, and Contrast Limited Adaptive Histogram Equalization (CLAHE) for image preprocessing. The model was trained on a curated subset of the Labeled Faces in the Wild (LFW) dataset, containing 20 categories with 50 images each, and evaluated using a locally captured dataset. Training was conducted on Google Colab using a pre-trained MobileNet model that was fine-tuned with 800 images, while 200 images were used for validation. System performance was assessed through several metrics, including accuracy, precision, recall, F1-score, and a confusion matrix. The model achieved a validation accuracy of 86% and an average F1-score of 0.85, reflecting high classification accuracy. To enhance usability, the system was implemented within a Python-based graphical user interface (GUI), which automates attendance tracking and records data directly into Excel spreadsheets. This study highlights the potential of integrating lightweight CNN architectures with effective preprocessing techniques and real-time GUI applications to create a reliable, efficient, and practical biometric attendance system