Deteksi Kantuk Menggunakan Ear Dan Mar Berbasis Deep Learning Pada Citra Sekuensial
Abstrak
Abstrak — Pada tahun 2023, kecelakaan lalu lintas mencapai angka tertinggi dalam lima tahun terakhir, dengan total 148.575 kasus, dimana 20% di antaranya disebabkan oleh kantuk, yang meningkatkan risiko kecelakaan hingga tiga kali lipat akibat penurunan kewaspadaan pengemudi. Penelitian ini mengusulkan metode ekstraksi fitur geometris Eye Aspect Ratio (EAR) dan Mouth Aspect Ratio (MAR) dari citra wajah sekuensial secara real-time menggunakan MediaPipe. EAR dan MAR dihitung berdasarkan koordinat landmark mata dan mulut, kemudian disusun dalam urutan temporal untuk menggambarkan perubahan kondisi subjek seiring waktu. Representasi ini efektif dalam menggambarkan transisi kantuk, yang dapat digunakan sebagai input dalam model deteksi berbasis deep learning. Penelitian ini melibatkan lima komponen utama: pengujian metode klasifikasi, pengolahan input data, perbaikan citra, augmentasi data, dan pipeline model. Data dari National Tsing Hua University Drowsiness Dataset (NTHU-DDD) dikelompokkan dalam window 60 frame, dengan fitur EAR dan MAR diekstraksi menggunakan MediaPipe. Hasil penelitian menunjukkan bahwa model CNN LSTM efektif dalam memproses fitur EAR dan MAR secara sekuensial. Representasi penuh dengan input implisit (120, 1) memberikan performa terbaik, sementara teknik augmentasi SMOTE meningkatkan performa dengan menyeimbangkan distribusi kelas. Model CNN-LSTM-120FT tanpa perbaikan citra atau augmentasi menunjukkan performa paling stabil, dengan accuracy 85,59% dan precision 92,31%.
Kata kunci— kecelakaan lalu lintas, kantuk, EAR, MAR, deep learning, citra sekuensial
Referensi
[1] A. D. Ayuningtyas, “2023 Capai Angka Tertinggi Kecelakaan Lalu Lintas 5 Tahun Terakhir,” goodstats.id. [Online]. Available: https://goodstats.id/article/2023-capai-angka-tertinggi-kecelakaan-lalu-lintas-5-tahun-terakhir-z67bk
[2] A. Moradi, S. S. H. Nazari, and K. Rahmani, “Sleepiness and the risk of road traffic accidents: A systematic review and meta-analysis of previous studies,” Transportation Research Part F: Traffic Psychology and Behaviour, vol. 65. pp. 620–629, 2019. doi: 10.1016/j.trf.2018.09.013.
[3] H. U. R. Siddiqui et al., “Non-Invasive Driver Drowsiness Detection System.” mdpi.com, 2021.
[4] D. Lestari, “KNKT: 80 Persen Kecelakaan di Tol Akibat Mengantuk dan Letih.” [Online]. Available: Siddqui, Hafeez Ur Rehman; Saleem, Adil Ali; Bbowrn, Robert; Bademci, Bahattin; Lee, Enesto; Rustam, Furqan; Dudley, Sandra
[5] D. C. S. Ram, D. J. V. . Koushik, and H. S. Pavan, “Drowsiness Detection using EAR (Eye Aspect Ratio) by Machine Learning,” Interantional Journal of Scientific Research in Engineering and Management, vol. 08, no. 01. pp. 1–13, 2024. doi: 10.55041/ijsrem19675.
[6] L.Thulasimani, Poojeevan P, Prithashasni S P, “Real Time Driver Drowsiness Detection using OpenCV and Facial Landmarks,” Journal of Contemporary Issues in Business and Government, vol. 27, no. 6. 2021. doi: 10.47750/cibg.2021.27.06.054.
[7] S. Plubin, W. Bunyatisai, M. Khamkong, T. Mouktonglang, and B. Plubin, “Multilabel-Thai Text Classification with Transformer-Rnn in Thai Banking Classification,” Pakistan Journal of Life and Social Sciences, vol. 23, no. 1. pp. 1–10, 2024. doi: 10.57239/PJLSS-2025-23.1.001.
[8] P. Inkeaw et al., “Automatic Driver Drowsiness Detection Using Artificial Neural Network Based on Visual Facial Descriptors: Pilot Study,” Nature and Science of Sleep, vol. 14. pp. 1641–1649, 2022. doi: 10.2147/NSS.S376755.
[9] J. Chandiwala and S. Agarwal, “Driver’s real-time Drowsiness Detection using Adaptable Eye Aspect Ratio and Smart Alarm System,” 2021 7th International Conference on Advanced Computing and Communication Systems, ICACCS 2021. pp. 1350–1355, 2021. doi: 10.1109/ICACCS51430.2021.9441756.
[10] R. Florez, F. Palomino-Quispe, R. J. Coaquira-Castillo, J. C. Herrera-Levano, T. Paixão, and A. B. Alvarez, “A CNN-Based Approach for Driver Drowsiness Detection by Real-Time Eye State Identification,” Applied Sciences (Switzerland), vol. 13, no. 13. 2023. doi: 10.3390/app13137849.
[11] S. Zu, Y. Jin, D. Yang, and H. Xu, “DrowsyNet: Multivariate Time Series Classification for Embedded Driver Drowsiness Detection,” 2022 8th International Conference on Control, Automation and Robotics, ICCAR 2022. pp. 442–451, 2022. doi: 10.1109/ICCAR55106.2022.9782626.
[12] F. Faraji, F. Lotfi, J. Khorramdel, A. Najafi, and A. Ghaffari, “Drowsiness Detection Based On Driver Temporal Behavior Using a New Developed Dataset.” 2021. [Online]. Available: http://arxiv.org/abs/2104.00125
[13] Saputra Desy, “Ratusan hektar tambak udang jebol akibat rob,” antaranews.com. [Online]. Available: https://www.antaranews.com/berita/289303/ratusan-hektar-tambak-udang-jebol-akibat-rob
[14] J. Wan, H. Liu, Y. Wu, Z. Lai, W. Min, and J. Liu, “Precise facial landmark detection by Dynamic Semantic Aggregation Transformer,” Pattern Recognition, vol. 156. 2024. doi: 10.1016/j.patcog.2024.110827.
[15] V. S. K. Rupani, V. V. S. Thushar, and K. Tejith, “Real-Time Drowsiness Detection Using Eye Aspect Ratio and Facial Landmark Detection,” arXiv preprint arXiv:2408.05836. 2024. [Online]. Available: https://arxiv.org/abs/2408.05836%0Ahttps://arxiv.org/pdf/2408.05836
[16] J. Holdsworth and M. Scapicchio, “What is deep learning?” [Online]. Available: https://www.ibm.com/topics/deep-learning
[17] M. Waqas and U. W. Humphries, “A Critical Review of RNN and LSTM Variants in Hydrological Time Series Predictions,” MethodsX. p. 102946, 2024.
[18] A. T. Mohan and D. V. Gaitonde, “A Deep Learning based Approach to Reduced Order Modeling for Turbulent Flow Control using LSTM Neural Networks,” no. April, 2018, doi: 10.48550/arXiv.1804.09269.
[19] A. M. Tahir et al., “Deep Learning for Reliable Classification of COVID-19, MERS, and SARS from Chest X-ray Images,” Cognitive Computation. 2022. doi: 10.1007/s12559-021-09955-1.
[20] S. Hao, D. H. Lee, and D. Zhao, “Sequence to sequence learning with attention mechanism for short-term passenger flow prediction in large-scale metro system,” Transp. Res. Part C Emerg. Technol., vol. 107, no. August, pp. 287–300, 2019, doi: 10.1016/j.trc.2019.08.005.
[21] A. Puente-Castro, E. Fernandez-Blanco, D. Rivero, and A. Molares-Ulloa, “Enhancing Orthopox Image Classification Using Hybrid Machine Learning and Deep Learning Models,” 2025, [Online]. Available: http://arxiv.org/abs/2506.06007
[22] Y. Dang et al., “Data Augmentation for Sequential Recommendation: A Survey,” 2024, [Online]. Available: http://arxiv.org/abs/2409.13545
[23] D. Lovell, D. Miller, J. Capra, and A. Bradley, “Never mind the metrics -- what about the uncertainty? Visualising confusion matrix metric distributions.” 2022. [Online]. Available: http://arxiv.org/abs/2206.02157
[24] J. Zou, C. Yuan, X. Zhang, G. Zou, and A. T. K. Wan, “Model Averaging for Support Vector Machine by Cross-Validation,” vol. 1, pp. 1–33, 2021, [Online]. Available: http://arxiv.org/abs/2112.14578
[25] F. Aldi, I. Nozomi, and S. Soeheri, “Comparison of Drug Type Classification Performance Using KNN Algorithm,” SinkrOn, vol. 7, no. 3, pp. 1028–1034, 2022, doi: 10.33395/sinkron.v7i3.11487.
[26] Muthukumar G. and Jyosna Philip, “CNN-LSTM Hybrid Deep Learning Model for Remaining Useful Life Estimation,” Int. J. Innov. Res. Multidiscip. F., vol. 10, no. Special Issue-54, pp. 38–50, 2024, [Online]. Available: https://www.ijirmf.com
[27] M. Cao, R. Yao, J. Xia, K. Jia, and H. Wang, “LSTM Attention Neural-Network-Based Signal Detection for Hybrid Modulated Faster-Than-Nyquist Optical Wireless Communications †,” Sensors, vol. 22, no. 22, 2022, doi: 10.3390/s22228992.
