Evaluating the Effect of Global Channel Attention on EfficientNetB0 for Distracted Driver Classification under Diverse Lighting Conditions

Zufar Abdullah Rabbani; Wahyu Syaifullah J. S; Alfan Rizaldy Pratama

doi:10.24252/instek.v11i2.67721

Authors

Zufar Abdullah Rabbani Universitas Pembangunan Nasional "Veteran" Jawa Timur
Wahyu Syaifullah J. S Universitas Pembangunan Nasional "Veteran" Jawa Timur
Alfan Rizaldy Pratama Universitas Pembangunan Nasional "Veteran" Jawa Timur

DOI:

https://doi.org/10.24252/instek.v11i2.67721

Keywords:

Channel Attention, EfficeintNetB0, Distracted Driver Classification

Abstract

Image-based distracted driver classification has been widely studied to reduce the risk of traffic accidents, yet most studies use data captured under uniform lighting conditions. However, illumination variation can degrade feature extraction in Convolutional Neural Network (CNN) architectures. Channel attention (CA) has previously been shown to improve EfficientNetB0, but EfficientNetB0 with CA has not yet been extensively tested for distracted driver classification under diverse lighting conditions. This study aims to evaluate the performance of channel attention on EfficientNetB0 in classifying distracted driver activities under diverse lighting conditions. We used the State Farm Distracted Driver Dataset and a primary dataset of 10.080 images from three subjects captured under dark, semi-bright, and bright lighting conditions. The secondary data and two primary subjects were used for training, while one unseen primary subject was used for testing. Preprocessing included image resizing, normalization, and data augmentation. The models were trained using two-phase transfer learning (frozen backbone followed by fine-tuning) and evaluated using accuracy, precision, recall, and F1-score. The accuracies of baseline EfficientNetB0 versus EfficientNetB0+CA were 38.84% vs 60.36% (dark), 61.07% vs 32.59% (semi-bright), 67.14% vs 84.73% (bright), and 55.68% vs 59.23% (all conditions). CA improved accuracy in dark and bright conditions but reduced it in semi-bright conditions, so a consistent performance gain from adding CA cannot be concluded.

Downloads

Download data is not yet available.

References

[1] D. Amalia and F. Utaminingrum, “Deteksi Kantuk pada Pengemudi melalui Jumlah Kedipan Mata Menggunakan Facial Landmark berbasis Intel NUC,” Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer, vol. 5, no. 12, pp. 5529–5535, 2021, [Online]. Available: http://j-ptiik.ub.ac.id

[2] H. Suseno, “Kecelakaan Menghambat Tumbuh dan Berkembang Suatu Bangsa,” Jurnal Indonesia Sosial Sains, vol. 3, no. 10, Oct. 2022, doi: https://doi.org/10.59141/jiss.v3i10.716.

[3] S. Firdaus and K. D. Artika, “DETEKSI KELELAHAN PENGEMUDI MOBIL MENGGUNAKAN CITRA WAJAH,” ELEMEN : JURNAL TEKNIK MESIN, vol. 8, no. 1, pp. 16–21, Jun. 2021, doi: 10.34128/je.v8i1.154.

[4] Armizoprades, A. Fadhli, A. Azhar, and T. Fauzan, “ANALYSIS OF FACTORS CAUSING TRAFFIC ACCIDENTS IN THE KM 45 SICINCIN AREA, PADANG PARIAMAN REGENCY,” Journal of Scientech Researchand Development, vol. 7, no. 1, Jun. 2025, doi: https://doi.org/10.56670/jsrd.v7i1.1161.

[5] G. Hermawan, “Klasifikasi Pengemudi Terganggu Berdasarkan Citra Menggunakan ResNet-50,” Jurnal Informatika Komputer, Bisnis dan Manajemen, vol. 22, no. 3, 2024, doi: https://doi.org/10.61805/fahma.v22i3.137.

[6] N. Darapaneni et al., “Distracted Driver Monitoring System Using AI,” in 2022 Interdisciplinary Research in Technology and Management (IRTM), IEEE, Feb. 2022, pp. 1–8. doi: 10.1109/IRTM54583.2022.9791545.

[7] M. Faizal Nazili, A. B. Firmansyah, and R. Purbaningtyas, “Klasifikasi Keparahan Demensia Alzheimer Menggunakan Metode Convolutional Neural Network pada Citra MRI Otak,” Indonesian Journal of Machine Learning and Computer Science, vol. 3, no. 1, pp. 1–7, 2023.

[8] X. Shi, F. Li, G. Lu, Y. Xie, F. Dong, and K. Chen, “A Recognition Algorithm for Distracted Driving Behavior Based on CBAM-EfficientNetB0,” in The 4th International Conference on Frontiers of Electronics, Information and Computation Technologies, 2025, pp. 127–138. doi: 10.1007/978-981-96-5318-8_12.

[9] H. Cai et al., “Enable Deep Learning on Mobile Devices: Methods, Systems, and Applications,” ACM Transact. Des. Autom. Electron. Syst., vol. 27, no. 3, pp. 1–50, May 2022, doi: 10.1145/3486618.

[10] T. Khan, G. Choi, and S. Lee, “EFFNet-CA: An Efficient Driver Distraction Detection Based on Multiscale Features Extractions and Channel Attention Mechanism,” Sensors, vol. 23, no. 8, Apr. 2023, doi: 10.3390/s23083835.

[11] Z. Zhao, “A comparative study of large-scale and lightweight convolutional neural networks for ImageNet classification,” Applied and Computational Engineering, vol. 47, no. 1, pp. 101–110, Mar. 2024, doi: 10.54254/2755-2721/47/20241236.

[12] D. Akhiyar, H. Herasmus, and Nofriadiman, “ADAPTASI SPESIFIK EFFICIENTNETB0 DENGAN LAPISAN KUSTOM UNTUK IDENTIFIKASI BUAH TROPIS,” Jurnal Sains Informatika Terapan, vol. 4, no. 2, pp. 308–313, 2025, Accessed: Sep. 14, 2025. [Online]. Available: https://rcf-indonesia.org/home/

[13] Sanjay Balwani and Narendra Bawane, “A Multiscale Fusion Network Integrating ConvNeXtSmall and EfficientNetB0 with Enhanced Image Preprocessing for Robust Classification,” Metallurgical and Materials Engineering, vol. 31, no. 3, pp. 382–391, Mar. 2025, doi: 10.63278/1388.

[14] M. Tan and Q. V. Le, “EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks,” Sep. 2020. [Online]. Available: http://arxiv.org/abs/1905.11946

[15] J. Hu, L. Shen, and G. Sun, “Squeeze-and-Excitation Networks,” in Conference on Computer Vision and Pattern Recognition (CVPR), IEEE/CVF, 2018, pp. 7132–7141. doi: 10.1109/CVPR.2018.00745.

[16] S. Bala et al., “Improving Kidney Stone Detection with YOLOV10 and Channel Attention Mechanisms in Medical Imaging,” Journal of Electronics, Electromedical Engineering, and Medical Informatics, vol. 7, no. 3, pp. 951–963, Jul. 2025, doi: 10.35882/jeeemi.v7i3.868.

[17] M. Uzzol Hossain, M. Ataur Rahman, M. Manowarul Islam, A. Akhter, M. Ashraf Uddin, and B. Kumar Paul, “Automatic driver distraction detection using deep convolutional neural networks,” Intelligent Systems with Applications, vol. 14, p. 75, 2022, doi: 10.1016/j.iswa.2022.20.

[18] J. Wang and Z. Wu, “Model Lightweighting for Real-time Distraction Detection on Resource-Limited Devices,” Comput. Intell. Neurosci., vol. 2022, pp. 1–13, Dec. 2022, doi: 10.1155/2022/7360170.

[19] Y. Lin, D. Cao, Z. Fu, Y. Huang, and Y. Song, “A Lightweight Attention-Based Network towards Distracted Driving Behavior Recognition,” Applied Sciences (Switzerland), vol. 12, no. 9, May 2022, doi: 10.3390/app12094191.

[20] L. Pasqualetto Cassinis, A. Menicucci, E. Gill, I. Ahrns, and M. Sanchez-Gestido, “On-ground validation of a CNN-based monocular pose estimation system for uncooperative spacecraft: Bridging domain shift in rendezvous scenarios,” Acta Astronaut., vol. 196, pp. 123–138, Jul. 2022, doi: 10.1016/j.actaastro.2022.04.002.

[21] I. Jegham, A. Ben Khalifa, I. Alouani, and M. A. Mahjoub, “Vision-based human action recognition: An overview and real world challenges,” Forensic Science International: Digital Investigation, vol. 32, p. 200901, Mar. 2020, doi: 10.1016/j.fsidi.2019.200901.

[22] J. Yang, H. Zou, Y. Zhou, and L. Xie, “Robust adversarial discriminative domain adaptation for real-world cross-domain visual recognition,” Neurocomputing, vol. 433, pp. 28–36, Apr. 2021, doi: 10.1016/j.neucom.2020.12.046.

[23] K. Chao, W. Song, S. Shao, D. Liu, X. Liu, and X. B. Zhao, “CUI-Net: a correcting uneven illumination net for low-light image enhancement,” Sci. Rep., vol. 13, no. 1, Dec. 2023, doi: 10.1038/s41598-023-39524-5.

[24] X. Niu and G. Zheng, “Research on Low-Light Image Enhancement Algorithm Based on Multi-Dimensional Attention,” Academic Journal of Engineering and Technology Science, vol. 8, no. 1, 2025, doi: 10.25236/ajets.2025.080109.

[25] A. Montoya, D. Holman, T. Smith, and W. Kan, “State Farm Distracted Driver Detection,” https://kaggle.com/competitions/state-farm-distracted-driver-detection.

[26] H. Talebi and P. Milanfar, “Learning to Resize Images for Computer Vision Tasks,” 2021.

[27] N. Fei, Y. Gao, Z. Lu, and T. Xiang, “Z-Score Normalization, Hubness, and Few-Shot Learning,” 2021.

[28] M. Kang, M. Kang, S. W. Lee, and S. Kim, “Mixup Mask Adaptation: Bridging the gap between input saliency and representations via attention mechanism in feature mixup,” Image Vis. Comput., vol. 146, Jun. 2024, doi: 10.1016/j.imavis.2024.105013.

[29] A. Howard et al., “Searching for MobileNetV3,” in International Conference on Computer Vision (ICCV), IEEE/CVF, 2019, pp. 1314–1324. doi: 10.1109/ICCV.2019.00140.

[30] N. Ma, X. Zhang, H.-T. Zheng, and J. Sun, “ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design,” 2018, pp. 122–138. doi: 10.1007/978-3-030-01264-9_8.

[31] Darmatasia and M. H. Hasrul, “Model Efisien Berbasis Mobile Vision Transformer (MobileViT) untuk Klasifikasi Jenis Tanah dari Citra,” JURNAL INSTEK, vol. 10, no. 2, pp. 518–533, Oct. 2025, [Online]. Available: https://journal.uin-alauddin.ac.id/index.php/instek