<![CDATA[Every institution such as the education system in Indonesia, even for offices certainly requires a system that can record the entire community of it's members. But in today's modern era, where technology has advanced rapidly it turns out that in some institutions in Indonesia still rely on the old presence system that is manually, such as using paper and initialing. This research makes it possible to reduce fraud by utilizing digital imagery that is face recognition in order to make a presence so that it becomes more practical, efficient, fast and certainly safe and does not happen to the detriment of any institution. In this study, a student presence system was developed based on face recognition using Local Binary Pattern and K-Nearest Neighbor method. By using the Logitech C270 webcam and the Alcatroz Stealth 5 mouse as an input producer, the Intel NUC5i7RYH Mini PC as the main processor, and 7-inch Waveshare monitor as output. Webcams produce images of students sitting in class and then processed by a Mini PC for the detection and facial recognition of each student. Obtained the names of facial recognition results that can enter the attendance list if the user (lecturer or researcher) presses the presence confirmation button on the application using the mouse. The average system accuracy of all experiments in face detection using Haar Cascade Classifier is 88.88%, in face recognition using Local Binary Pattern and K-Nearest Neighbor for k = 3 value is 78.125%, for k = 5 value is 74.375%, and for the value of k = 7 which is 68.125% so that the highest accuracy can be achieved using the value k = 3. The average computational time of all experiments in face detection is 26.2 ms while for face recognition is 371.675 ms.]]>
Copyrights © 2020
