This project is an image-based drowsiness detection system designed to detect whether a driver is drowsy or alert using computer vision and deep learning methods. It aims to reduce traffic accidents caused by driver fatigue by providing an early warning system based on image classification.
In Indonesia, approximately 80% of highway accidents are caused by drowsy or fatigued drivers. From October to December 2019 alone, more than 3,500 accidents were related to drowsiness—accounting for 9.5% of all traffic accidents during that period. This project is a step toward building an effective system that can help prevent such incidents.
| Model | Accuracy |
|---|---|
| EAR + MAR | 0.64 |
| EAR + MAR + SVM | 0.69 |
| CNN | 0.87 |
| VGG16 (Transfer Learning) | 0.71 |
| ResNet50 (Transfer Learning) | 0.51 |
📌 CNN outperformed other methods with the highest accuracy of 87%.