Deteksi Kedipan dengan Metode CNN dan Percentage of Eyelid Closure (PERCLOS)

Septiandi, Lutfi Ananditya (2021) Deteksi Kedipan dengan Metode CNN dan Percentage of Eyelid Closure (PERCLOS). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
1. Buku TA_Lutfi Ananditya Septiandi_07211640000057.pdf
Restricted to Repository staff only

Download (7MB) | Request a copy

Abstract

Pengembangan teknologi mengenai face detection dan eyes detection melaju sangat pesat, sehingga peneliti berlomba-lomba meneliti metode dan algoritma yang optimal untuk pengaplikasian di kehidupan sehari-hari, mulai dari pengamanan biometrics sampai identifikasi wajah secara au- tomasi. Di tugas akhir ini diusulkan penggunaan metode Convo- lutional Neural Network (CNN) dan Percentage of Eyelid Closure (PERCLOS) pada pendeteksi kedipan mata. Sistem dibangun menggunakan webcam personal computer sebagai kamera dan mendeteksi secara real-time. Sistem dapat mengenali kondisi ketika mata tertutup atau mata terbuka dan menentukan lebar bukaan mata dengan menggunakan Eye Aspect Ratio (EAR) serta dapat mengestimasi skor tatapan dengan menggunakan Percentage of Eyelid Closure (PERCLOS). Sistem dapat menge- nali wajah dari objek bukan wajah dengan jarak pendeteksian optimal antara 40-70 cm. Model hasil training dapat mengk- lasifikasikan kondisi mata terbuka dan mata tertutup dengan menggunakan Convolutional Neural Network dengan arstitektur yang memiliki 3 layer mendapatkan hasil accuracy 98% ================================================================================================= The development of technology on face detection and eyes detection is progressing very rapidly, so researchers are vying to research opti- mal methods and algorithms for application in everyday life, ranging from biometrics security to automatic facial identi�cation. In this �nal projects is proposed the use of Convolutional Neural Network (CNN) and Percentage of Eyelid Closure (PERCLOS) methods on eye-blink detection. The systems is built using personal computer webcam as a camera and detecting in real-time. The systems can recognize conditions when eyes are closed or eyes open and determi- ne the width of the eyes openings by using Eye Aspect Ratio (EAR) and can estimate stare scores by using Percentage of Eyelid Closure (PERCLOS). The system can recognize faces from non-face obje- cts with an optimal detection distance between 40 until 70 cm. The training result model can classify the conditions of open and closed eyes by using a Convolutional Neural Network with an architecture that has 3 layers to get an accuracy of 98%

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Face detection, Eyes detection, Convolutional Neural Network (CNN), Eye Aspect Ratio (EAR), Percentage of Eyelid Closure (PERCLOS). Face detection, Eyes detection, Convolutional Neural Network (CNN), Eye Aspect Ratio (EAR), Percentage of Eyelid Closure (PERCLOS).
Subjects: T Technology > T Technology (General)
T Technology > T Technology (General) > T11 Technical writing. Scientific Writing
T Technology > T Technology (General) > T57.5 Data Processing
T Technology > T Technology (General) > T57.62 Simulation
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User: Septiandi Lutfi Ananditya
Date Deposited: 10 Mar 2021 06:14
Last Modified: 10 Mar 2021 06:14
URI: https://repository.its.ac.id/id/eprint/84056

Actions (login required)

View Item View Item