Deteksi Kantuk Berdasarkan Pola Kedipan Mata Dengan Metode Gated Recurrent Unit Pada Mobile Device

Atho, M. Radhito (2025) Deteksi Kantuk Berdasarkan Pola Kedipan Mata Dengan Metode Gated Recurrent Unit Pada Mobile Device. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024211056-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Setiap tahunnya, kecelakaan lalu lintas akibat pengemudi yang tertidur saat berkendara menjadi penyebab utama tingginya angka korban jiwa di Indonesia, terutama pada pengguna sepeda motor. Penelitian ini bertujuan untuk mengembangkan sistem deteksi kantuk berbasis pembelajaran mesin guna meminimalkan risiko tersebut. Proses deteksi dilakukan dengan dua tahap utama: pertama, klasifikasi kondisi mata (terbuka atau tertutup) menggunakan model Convolutional Neural Network (CNN) yang diimplementasikan pada perangkat OpenMV; kedua, pemodelan pola kedipan mata dalam bentuk data sekuensial dengan memanfaatkan Gated Recurrent Unit (GRU) untuk mengklasifikasikan tingkat kantuk pengguna. Data pelatihan berasal dari dataset DROZY yang telah diproses menjadi urutan biner, sementara data pengujian juga mencakup dataset UTA-RLDD dan pengujian langsung pada pengguna. Sistem ini dirancang untuk berjalan secara real-time pada perangkat Android dan menunjukkan performa baik dalam klasifikasi tiga tingkat kantuk: terjaga, transisi, dan mengantuk. Hasil penelitian menunjukkan bahwa sistem yang dibangun mampu beroperasi secara efisien di perangkat dengan sumber daya terbatas dan dapat berkontribusi sebagai solusi praktis dalam meningkatkan keselamatan berkendara.
=================================================================================================================================
Every year, traffic accidents due to drivers falling asleep while driving are the main cause of high fatalities in Indonesia, especially for motorcycle users. This study aims to develop a machine learning-based drowsiness detection system to minimize this risk. The detection process is carried out in two main stages: first, classification of eye conditions (open or closed) using the Convolutional Neural Network (CNN) model implemented on the OpenMV device; second, modeling eye blink patterns in the form of sequential data by utilizing Gated Recurrent Uni (GRU) to classify the user’s level of drowsiness. The training data comes from the DROZY dataset that has been processed into a binary sequence, while the testing data also includes the UTA-RLDD dataset and direct testing on users. This system is designed to run in real time on Android devices and shows good performance in classifying three levels of drowsiness: awake, transitional, and drowsy. The results of the study show that the system built is able to operate efficiently on devices with limited resources and can contribute as a practical solution in improving driving safety.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Deteksi kantuk, pola kedipan mata, Gated Recurrent Unit (GRU), perangkat mobile, Convolutional Neural Network (CNN), Drowsiness detection, eye blink pattern, Gated Recurrent Unit (GRU), mobile device, Convolutional Neural Network (CNN)
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HE Transportation and Communications > HE5614.2 Traffic safety Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.774.A53 Android Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) R Medicine > R Medicine (General) > R858 Deep Learning T Technology > T Technology (General) > T57.5 Data Processing T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.585 TCP/IP (Computer network protocol)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	M. Radhito Bil Atho
Date Deposited:	30 Jul 2025 01:44
Last Modified:	30 Jul 2025 01:44
URI:	http://repository.its.ac.id/id/eprint/123094

Actions (login required)

View Item