Driver-drowsiness-detection

YOLOv8 Driver drowsiness detection Deep learning project Enhanced by algorithm based on Facial Features

purpose

Receive driver's eye status data using a YOLOv8 model. If the eyes are closed for 70% of frames within a 4.5-second window, trigger a sound alarm. If the eyes are open for 0.5 seconds, turn off the alarm and consider it not drowsy.

operating principle

In Main.py, create and run four multiprocesses via multiprocessing:

• Drowsiness detection algorithm
• FPS and time calculation algorithm
• OpenCV and model inference algorithm
• process infrenced result and show true imshow function Control these processes through a GUI.

models

optimization & performance

data

source	purpose	link
Driver and Passenger State and Anomaly Behavior Monitoring Data from Aihub	detect face eye mouth	https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=data&dataSetSn=173

ppt

google drive link: 🔗

Presentation Format and Compatibility

• Microsoft PowerPoint Format: This presentation was created in Microsoft PowerPoint for optimal viewing.
• Potential Animation Issues: Please be aware that certain animations may not function properly when using other presentation software.
• Recommendation: For the best viewing experience, it is recommended to use Microsoft PowerPoint to open and present this file.

environment

1. windows *linux env not runnable winsoud is used 🔗, which is win32api
2. numpy
3. opencv
4. ipywidgets
5. pytorch
6. openvino (in case cuda unavailable)