rki.news
Source: Xinhua
SYDNEY, March 11: Australian researchers have developed an artificial intelligence (AI) system capable of detecting dangerous driving behaviors through facial analysis, offering a potential non-invasive alternative to traditional breathalyzer tests.
According to a statement from Edith Cowan University (ECU), the new technology uses advanced 3D facial analysis to identify three key risk factors linked to road accidents: alcohol intoxication, driver fatigue and emotional expressions such as anger.
The system, created by ECU researchers, operates through a single deep-learning model that can detect blood alcohol concentration with nearly 90 percent accuracy and identify driver drowsiness with about 95 percent accuracy.
Researchers said the technology can also determine levels of intoxication by classifying alcohol impairment into three categories: sober, moderate and severe.
Lead researcher and ECU PhD candidate Abdullah Tariq explained that the system automatically analyzes various facial dynamics, including eye blinking patterns, subtle facial movements and progressive changes in facial features. These indicators help distinguish between different physical and emotional conditions affecting drivers.
Unlike conventional breathalyzers, the AI-based approach enables continuous, real-time monitoring without requiring active cooperation from the driver.
A companion study further showed that combining infrared and color video technology improves detection accuracy in low-light environments.
Researchers believe the innovation could provide authorities with a modern, non-intrusive tool to help reduce drink driving and improve road safety.