Every year, alcohol levels wreak havoc on the roads and despite awareness campaigns, there are still serious and even fatal accidents. To combat this, researchers at Edith Cowan University in Australia have designed a technology to detect drunk drivers.
75% drunk level accuracy
To do this, their technology scans video frames captured by so-called cameras “intelligent”. The researchers conducted their experiment on three groups with three blood alcohol levels in a controlled environment but “realistic”. When individuals drove in a simulator, the technology scanned images of faces captured on video using current RGB standards.
Advertisement, your content continues below
Then, facial expressions, gaze direction and head position allowed their technology to detect intoxication in these drivers. And even more since researchers are able to know the blood alcohol level with an announced accuracy of 75% on the three levels, according to them.
The researchers say the low-resolution images are enough for their technology to work and that it could be used by roadside surveillance cameras or by law enforcement.
Limits to this detection technology
However, researchers recognize the limits of this technology since the drunk driver must be in his vehicle, therefore driving and endangering his life in addition to that of other road users. Their tool could be strengthened in the event of large-scale distribution by being combined with other solutions to identify motorists who are intoxicated.
Advertisement, your content continues below
