Roads To Be More Safer Thanks To Emotion Detectors

It has not been long since the news of development of an #-Link-Snipped-# hit the headlines and the rapid developments in AI have not escaped the field of automation safety. It is a known fact that along with fatigue and intoxication, the other main reason behind accidents involving human errors is the driver's mental state. Now a technology has come up that allows us to read a driver's facial expressions resulting in safer roads. The researchers at EPFL, Switzerland teamed up with PSA Peugeot Citroën and have now developed an on-board emotion detector based on the analysis of facial expressions.

The idea of reading a person's facial expressions is not exactly new but it was difficult to achieve this using non-invasive techniques. The EPFL Signal Processing Lab 5, which specialises in the facial detection, it's monitoring and analysis domain, placed infrared cameras behind the steering wheel of the car. Now the main problem was- How to detect the mood of a person if each one of us has a different way of expressing a similar disposition?


To start with, the research chiefs Hua Gao and Anil Yüce chose to work only with two expressions: anger and disgust (which manifests similar to anger). In the first phase of tests, the system learned to 'identify' the expressions of the subjects based on a series of photos which showed the subjects expressing angner or disgust. In the second phase, the system was required to detect the same expressions from a video.

The results from the tests were satisfactory. The display of 'irritation' was detected correctly by the system most of the times. In cases where the system failed, Hua Gao commented that it happened due to the variety of outcomes that vary from individual to individual. There will always be new challenges in this part, Gao added. Additional research will focus on updating the database in real time using a smart human-machine interface or using a more advanced facial monitoring algorithm.

Things will go beyond that, too. In this very project, the emotion detector was coupled with a fatigue detector which measured the percentage of eyelid closure. Further projects are aimed at lip reading for use in vocal recognition and on detecting other emotions such as distraction and sadness for maximum driver comfort and safety.

Source: Emotion detectors could make driving safer - EPFL