New camera technology detects drunk drivers based on facial features

New camera technology detects drunk drivers based on facial features
New camera technology detects drunk drivers based on facial features


Image credit: Pixabay/CC0 Public Domain

Researchers at Edith Cowan University (ECU) are developing a new computer tracking technology that uses data from camera footage to determine whether or not a driver is under the influence of alcohol when they get behind the wheel.

In collaboration with Mix by Powerfleet, scientists collected data from drunk drivers in a controlled but realistic environment and shot a series of videos of the drivers.

Participants, who were at three levels of alcohol consumption – sober, slightly intoxicated and heavily intoxicated – were recorded driving in a simulator.

The researchers then presented a machine learning system that uses recognizable cues from standard RGB (red, green, and blue) videos of the driver’s face to determine the level of alcohol impairment, including facial features, gaze direction, and head position. The research was presented at the IEEE/CVF Winter Conference on Applications of Computer Vision.

“Our system detects different levels of impairment caused by alcohol intoxication with an overall accuracy of 75% in the three-stage classification,” said ECU doctoral student Ensiyeh Keshtkaran.

“This will not only benefit vehicles equipped with driver monitoring systems and eye-tracking technologies, but may also be extended to smartphones, making the detection of alcohol intoxication more effective.

“Our system is able to detect the level of intoxication at the start of a journey, potentially preventing drunk drivers from hitting the road. This distinguishes it from methods based on observable driving behavior that require prolonged active vehicle operation to detect impairment.”

According to Dr. Syed Zulqarnain Gilani, a lecturer at ECU, the new technology is the first to use a standard RGB camera to determine the level of alcohol intoxication based on signs of impairment on the driver’s face.

“This research confirms that it is possible to determine the level of drunkenness using a simple camera. The next step in our research is to define the image resolution required to use this algorithm. If low-resolution videos prove sufficient, this technology can be deployed by roadside surveillance cameras and law enforcement agencies can use it to prevent (drunk) driving.”

A computer vision-based approach could potentially be integrated into road cameras in the future, similar to how these cameras currently detect seatbelt use or mobile phone activity, making it applicable to different types of vehicles without the need for special installations in the cabin.

The technology also includes 3D and infrared videos of the driver’s face, RGB rearview videos showing the driver’s posture and steering interactions, driving simulation event logs and screen recordings of driving behavior.

“The availability of this dataset not only enriches our research efforts but also provides the broader scientific community with an invaluable resource for further investigation and study,” explained Dr. Gilani.

Drunk driver detection – the future

Drink driving is the main factor in around 30% of all fatal road accidents in Australia. The Transport Accident Commission notes that one in five drivers killed on Australian roads have a blood alcohol content (BAC) of 0.05 or more.

“Existing approaches to detecting drunk drivers, which rely primarily on random breathalyzer tests, do not address this pressing problem,” said Ms Keshtkaran.

“Although work is underway to integrate drink-driving detection systems into future generations of vehicles and the introduction of autonomous vehicles is in sight, the ongoing problem of drink-driving remains a pressing concern.”

Ms. Keshtkaran noted that current research in the area of ​​drunk driving detection mainly focuses on analyzing driving behavior, such as driving and steering patterns, pedal usage, and vehicle speed. Some other approaches involve external sensors such as alcohol detection or touch-based sensors.

However, the potential of using computer vision techniques to detect noise phenomena based on biobehavioral changes in drivers has been very poorly explored.

“A key limitation of using driving behavior to detect drunk driving is the requirement that the driver must be actively controlling the vehicle for an extended period of time before their behavior can be assessed and identified as indicative of drunkenness. This implies that a potentially impaired driver is already on the road, endangering themselves and other road users. Rapid detection is critical to identifying drunk drivers and preventing them from endangering public safety,” she said.

More information:
Ensiyeh Keshtkaran et al, Estimation of blood alcohol level using facial features to assess driving impairment, 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) (2024). DOI: 10.1109/WACV57701.2024.00448

Provided by Edith Cowan University

Quote: New camera technology uses facial expressions to detect drunk drivers (June 24, 2024), accessed June 24, 2024 from

This document is subject to copyright. Except for the purposes of private study or research, no part of it may be reproduced without written permission. The contents are for information purposes only.

Leave a Reply

Your email address will not be published. Required fields are marked *