Automatic Presence Design in Virtual Class Using Convolution Neuron Network (CNN)

Abstract

Universitas Terbuka is a pioneer in distance learning, it is required to always innovate, especially in the mode of learning. One of learning is through virtual classes, the tutor as a teacher provides direction and assistance to students as participants. Tutor and student attendance can be written directly by the tutor or through the attendance link, this tends to have a lack of human error, either the tutor forgets to write down the participants who were present or the attendance link filled in by participants who were not present, because the link can be accessed by anyone. Based on this, we try to design an automatic presence that is carried out in a virtual class. The research result design of image database, design of user interface, and design of learning engine. The results of database testing show that every table in the database is normal and there is a logical relationship between tables. The user interface test results are proven to be able to accommodate the database and engine display deep learning method. The results of the learning engine show the creation of a deep learning method architecture in accordance with the needs of face recognition in the virtual class.