Driver drowsiness has become one of the huge factors resulting in vehicle accidents. According to a study, 21% of all fatal motorway accidents are caused due to heavy drowsy driving and around 1,00,000 accidents are the result of driver drowsiness every year. A number of lives can be saved if the drowsy driver is alerted in time. Our work analyses previous techniques and their limitations and proposes a system developed to detect and alert a drowsy driver which could solve previous problems. The system works in a non-intrusive manner and hence is very practical and feasible to implement. A webcam or a personal mobile phone is used to continuously send the driver feed to a Machine Learning model which will then with the help of technologies like OpenCV and Machine Learning techniques like HAAR cascading and facial landmarks detection look for drowsiness in real-time. This system will closely monitor facial features like blinking frequency, blinking duration, and yawning frequency which are very common factors that give away drowsiness. Once drowsiness is detected the system will raise an alarm depending on the level of drowsiness.
-
Non-Intrusive: No physical device is required to be attached to the human body
-
Light weight: Can be easily implemented in any smartphone or camera device.
-
Uses low-cost equipment: Using just a smartphone, hence no extra additional device cost is involved
-
Robustness: Works accurately in low light conditions and various face angles
-
Fast to adapt: No former machine learning is required and is ready to use directly.
This system can easily be implemented in any device with low technical requirements (any device with camera). Using this system, we can count the number of times the driver blinked & yawned. It also accurately calculates the percentage of eyes opened. This can make the system easy to predict drowsiness. If the system detects drowsiness, it will quickly ring an alarm using the speakers in the vehicle making the driver aware to take a break and avoid any kind of accidents that may occur due to drowsiness.
Android APK: download
Deployed at: drowsi-40aa2.web.app
Distributed under the Apache License. See LICENSE for more information.
-
Hevesh Lakhwani : @HeveshL - hevesh000@gmail.com
-
Om Jodhpurkar : @infinity811 - ojodhpurkar@gmail.com
-
Pranav Bhagat : @prnv30 - pranavbs30@gmail.com
-
Saquib Sheikh : @saquib-alpha - saquibsheikh2000@gmail.com