Deteksi dan Pengenalan Wajah Real-time Menggunakan Transfer Learning pada Arsitektur VGG16 Pada Event Skala Besar

Abstract

The development of artificial intelligence technology has driven increased efficiency across various sectors, including the management of large-scale events such as concerts, football matches, and wedding ceremonies. One of the common issues in these events is the attendance checking process, which is still conducted manually, leading to queues, delays, and potential identification errors. This study proposes a real-time face detection and recognition system using a transfer learning approach on the VGG16 architecture, a high-performance Convolutional Neural Network (CNN) model. The system is designed to recognize participants' faces directly through a camera upon arrival, without requiring physical interaction or manual matching. The facial dataset was collected prior to the event and used to train the model pre-trained on VGGFace. Testing results show that the system can recognize participant faces with an accuracy up to 99.5% during training and 100% during validation. and operates in real-time with fast response under various lighting conditions and facial angles. This research contributes to the development of an automated face-based attendance system, supporting the implementation of a smart event system as part of a broader smart city concept.