Team

Francisco Lozano

Description

The objective of this project is to develop and evaluate an automated driver drowsiness detection system using facial analysis on one frame. This project aims to enhance road safety by identifying drowsy driving behavior and alerting the driver. Using image processing we can accurately detect signs of drowsiness based on facial features, such as eye closure duration. Most driver drowsiness detection systems rely on video; using a single image for classification could speed up computation.

Directory setup

• data: folder that contains the raw & preprocessed data for the project
• output: contains outputs such as data/graphs needed for the project - these outputs are generated by the jupyter notebooks
• app: contains the source code for the gradio app in HuggingFace
• images: containes images used by notebooks for reporting (This DOES NOT include images used in the algorithm)
• models: contains models used by the notebooks.
• present: contains my presentation material.

Dataset

https://www.kaggle.com/datasets/ismailnasri20/driver-drowsiness-dataset-ddd

App Deployment

I deployed my final model in HuggingFace via a gradio app. Check it out! TODO

References:

• A Review of Recent Developments in Driver Drowsiness Detection Systems, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8914892/
• Driver drowsiness monitoring system using visual behaviour and machine learning, https://ieeexplore.ieee.org/document/8405495
• Drivers Drowsiness Detection using Image Processing and I-Ear Techniques, https://ieeexplore.ieee.org/document/10142501
• X. Guo, "LIME: A Method for Low-light Image Enhancement," Proceedings of the 24th ACM International Conference on Multimedia, Amsterdam, The Netherlands, 2016, pp. 87-91, doi: 10.1145/2964284.2967188. https://doi.org/10.1145/2964284.2967188
• Fei Ni, Zhuang Fu, QiXin Cao, YanZheng Zhao. “Image processing method for eyes location based on segmentation texture”. Sensors and Actuators A: Physical, Volume 143, Issue 2, 2008, Pages 439-451. ISSN 0924-4247. https://doi.org/10.1016/j.sna.2007.11.033.
• Albadawi Y, Takruri M, Awad M. “A Review of Recent Developments in Driver Drowsiness Detection Systems”. Sensors (Basel). 2022 Mar 7;22(5):2069. doi: 10.3390/s22052069. PMID: 35271215; PMCID: PMC8914892.
• M. A. Faidhi Daud, A. P. Ismail, N. M. Tahir, K. Daud, N. M. Kasim and F. A. Mohamad, "Real Time Drowsy Driver Detection Using Image Processing on Python," 2022 IEEE 12th International Conference on Control System, Computing and Engineering (ICCSCE), Penang, Malaysia, 2022, pp. 131-136, doi: 10.1109/ICCSCE54767.2022.9935627.
• A. Kumar and R. Patra, "Driver drowsiness monitoring system using visual behaviour and machine learning," 2018 IEEE Symposium on Computer Applications & Industrial Electronics (ISCAIE), Penang, Malaysia, 2018, pp. 339-344, doi: 10.1109/ISCAIE.2018.8405495.
• Nasri, I., Karrouchi, M., Snoussi, H., Kassmi, K., Messaoudi, A. (2022). Detection and Prediction of Driver Drowsiness for the Prevention of Road Accidents Using Deep Neural Networks Techniques. In: Bennani, S., Lakhrissi, Y., Khaissidi, G., Mansouri, A., Khamlichi, Y. (eds) WITS 2020. Lecture Notes in Electrical Engineering, vol 745. Springer, Singapore. https://doi.org/10.1007/978-981-33-6893-4_6
• Kazemi, V., & Sullivan, J. (2014). One millisecond face alignment with an ensemble of regression trees. 2014 IEEE Conference on Computer Vision and Pattern Recognition, 1867-1874. https://www.semanticscholar.org/paper/One-millisecond-face-alignment-with-an-ensemble-of-Kazemi-Sullivan/d78b6a5b0dcaa81b1faea5fb0000045a62513567?p2df

TODO List

1. Build a gradio app implementing the system for a better UI/UX experience.
   • Just make it analyze images, no video stream