Sign In

Communications of the ACM

ACM TechNews

Self-Driving Cars Risk 'Future Errors' Due to Difficulty Detecting Darker Skin Tones


Pedestrians.

Georgia Institute of Technology researchers have found that state-of-the-art object-detection systems are better at detecting people with lighter skin tones, meaning they are less likely to identify black pedestrians and to stop before crashing into them.

Credit: Wikimedia Commons

Researchers at the Georgia Institute of Technology (Georgia Tech) have found that state-of-the-art object-detection systems, such as the sensors and cameras used in self-driving cars, are better at detecting people with lighter skin tones, meaning they are less likely to identify black people and to stop before crashing into them.

The researchers examined eight image recognition systems and found the bias in each one, with accuracy 5% lower on average for people with darker skin.

The team proved the hypothesis by dividing a large pool of pedestrian images into groups of lighter and darker skin using the Fitzpatrick scale—a scientific way of classifying skin color.

“This behavior suggests that future errors made by autonomous vehicles may not be evenly distributed across different demographic groups,” the researchers wrote.

From The Washington Times
View Full Article

 

Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA


 

No entries found