Dataset: http://mrl.cs.vsb.cz/eyedataset
Haarcascade: https://drive.google.com/file/d/1TedeO86Tyl81PsNwxEJrJf7UFnBk3S31/view?usp=sharing
Problem Statement: Driving as an activity is tiresome after a few hours of driving and especially when you are returning home after a long day of work. In such cases, tiredness induces sleeping in the driver. A study conducted in 2020, revealed that 50% of the truck drivers felt sleepy or tiresome while driving. Such drowsy drivers are 2.6 times more prone to an accident than general drivers. In 2016, the World Bank conducted a study that deaths and injuries caused by road crashes cost around 7.5% of India’s GDP.
Solution Implemented: This project is a small initiative to apply Artificial Intelligence as a boon to save lives and promote safe driving practices. It focuses on making use of Machine Learning in identifying whether a driver is about to sleep with the “blinking eyes actions” and switching on an alarm system to make the driver alert of his sleepiness. This would help the driver in a sudden waking up action. Furthermore, the driver could take a wise and rational decision to stop and rest and prevent the destruction of life and property. To reduce the risk of accidents, an alarm is set for the driver in Automobiles, whenever he/she is about to sleep for a specified minimum amount of time. Data preprocessing was performed based on the annotation of the images. TensorFlow, CNN, Deep Learning, ImageNet have been used to detect driver drowsiness. The Pygame module is also used to set an alarm sound for the driver.
Accuracy Score: Test Data : 95.185%
The Demo-Video of the result and proper Presentation of the project is here. : https://www.canva.com/design/DAErjmL9-q8/neADkW0wH8u6Wq-sC4q46g/view?utm_content=DAErjmL9-q8&utm_campaign=designshare&utm_medium=link&utm_source=publishsharelink
Future Scope of the Project:
- volume modulation so that the driver doesn't abruptly wake up and chances of sudden vehicle disbalance reduces.
- Signaling the traffic when the driver is detected to be asleep or about to be asleep through blinking lights in an unique pattern, this will wARn the other drivers and further reduce accidental probabilities WITH THE HELP OF IOT.