Facial recognition software may omit transgenders

Facial recognition software can categorise the gender of many men and women with remarkable accuracy but if that face belongs to a transgender person, such systems get it wrong more than one third of the time, says a new study.

“We found that facial analysis services performed consistently worse on transgender individuals, and were universally unable to classify non-binary genders," said study lead author Morgan Klaus Scheuerman from the University of Colorado Boulder. "While there are many different types of people out there, these systems have an extremely limited view of what gender looks like,” Scheuerman added. Previous research suggests they tend to be most accurate when assessing the gender of white men, but misidentify women of colour as much as one-third of the time. “We knew there were inherent biases in these systems around race and ethnicity and we suspected there would also be problems around gender,” said senior author Jed Brubaker. “We set out to test this in the real world,” Brubaker said.

For the findings, researchers collected 2,450 images of faces from Instagram, each of which had been labelled by its owner with a hashtag indicating their gender identity. The pictures were then divided into seven groups of 350 images (#women, #man, #transwoman, #transman, #agender, #agenderqueer, #nonbinary) and analysed by four of the largest providers of facial analysis services (IBM, Amazon, Microsoft and Clarifai).

Notably, Google was not included because it does not offer gender recognition services. On average, the systems were most accurate with photos of cisgender women (those born female and identifying as female), getting their gender right 98.3 per cent of the time. They categorised cisgender men accurately 97.6 per cent of the time. But trans men were wrongly identified as women up to 38 per cent of the time. And those who identified as agender, genderqueer or nonbinary —indicating that they identify as neither male or female — were mischaracterised 100 per cent of the time.