New AI-Driven Camera Technology Detects Drunk Drivers
Researchers at Edith Cowan University (ECU) are developing a new technology that uses camera footage to detect if a driver is impaired by alcohol. In collaboration with Mix by Powerfleet, they collected data from drivers with varying levels of intoxication using driving simulators. The system utilizes machine learning to analyze facial features, gaze direction, and head position from standard RGB videos to determine alcohol impairment with 75% accuracy.
This technology aims to prevent impaired driving by detecting intoxication levels before driving begins, unlike current methods that rely on observable driving behaviors. The research, presented at the IEEE/CVF Winter Conference on Applications of Computer Vision, shows potential for integration into roadside surveillance cameras, enhancing law enforcement efforts.
The approach could extend beyond vehicles to smartphones, improving alcohol intoxication detection. ECU’s dataset, including 3D and infrared videos, driving simulation logs, and screen recordings, offers valuable resources for further research. With drunk driving contributing to 30% of fatal crashes in Australia, this innovation could address a significant public safety issue.