The advancement of technology has brought about a new breakthrough in vehicle safety with the development of an AI algorithm that can detect drunk drivers by scanning their faces. This innovative project, presented in a paper at an IEEE and CVF conference, aims to reduce drunk driving accidents by allowing in-car computing systems to assess the driver’s intoxication level as soon as they enter the vehicle with an impressive accuracy rate of 75%.
Unlike current methods that rely on observable behaviors like steering patterns and pedal usage, this new system uses a single color camera to monitor variables such as gaze direction and head position. Additionally, it can incorporate 3D and infrared footage of the driver’s face, rearview videos, steering interactions, event logs, and screen recordings of driving behavior to make a comprehensive assessment.
Ensiyeh Keshtkaran, a doctoral student at Edith Cowan University, Australia, involved in the project, highlighted the potential of this software to prevent impaired drivers from getting on the road by identifying intoxication levels at the beginning of a drive. With the ability to seamlessly integrate into smart vehicle architectures, such as eye tracking and driver monitoring systems, this technology could easily transition to other environments like smartphones.
The World Health Organization reports that alcohol impairment is a significant factor in 20% to 30% of fatal car accidents globally, with Australia seeing 30% of fatal crashes involving drivers exceeding the legal blood alcohol limit. Despite efforts to implement driver alcohol detection systems in future vehicles and the rise of autonomous cars, drunk driving remains a pressing issue that requires urgent attention.
The study conducted by the research team involved analyzing video footage of drivers of varying ages, drinking habits, and driving experience under different levels of intoxication. By collaborating with software company MiX by Powerfleet, they were able to identify visual cues of intoxication in the video footage and successfully predict a driver’s state in 75% of cases. These cues include bloodshot eyes, flushed face, droopy eyelids, and a dazed look.
Project lead Syed Zulqarnain Gilani expressed the importance of enhancing the image resolution data for the algorithm to improve its accuracy further. The ultimate goal is for this technology to be utilized by surveillance cameras on roadsides to prevent drunk driving incidents. This groundbreaking development marks a significant advancement as it can detect intoxication levels even before the vehicle is in motion, potentially preventing accidents and saving lives in the future.
In conclusion, the integration of AI algorithms into vehicle safety systems represents a promising step towards reducing the risks associated with drunk driving. By leveraging facial recognition technology and advanced computing systems, we may soon see a future where smart cars can prevent intoxicated individuals from operating vehicles, ultimately making roads safer for everyone.
