Google Developing Own Skin-Tone System In Diversity Push


Google develops its own alternative to industry-standard skin-tone classification system as it seeks to ensure products work well across skin types

Google is developing an alternative to the industry-standard system for classifying skin tones, as it seeks to improve inclusiveness, Reuters reported.

The Fitzpatrick Skin Type (FST) system has been in use by dermatologists since the 1970s and is now used by tech companies to test that products such as facial recognition systems or smartwatch heart-rate senors work equally well across different skin tones.

But many say the six-colour scale, which has four categories for white skin and one each for brown and black, is inadequate.

For instance, researchers at the US Department of Homeland Security last year recommended that FST be abandoned for evaluating facial recognition because it poorly represents colour range in diverse populations.

Intelligent Spaces: Using Smart Cities

Alternative system

Google said it is working on “more inclusive” measures as an alternative to FST.

“We are working on alternative, more inclusive, measures that could be useful in the development of our products, and will collaborate with scientific and medical experts, as well as groups working with communities of colour,” the company told Reuters, without giving further details.

The controversy around FST is part of a broader pressure on the tech industry to increase the diversity of its workforce.

It also has practical applications as companies look to assure users their products will work regardless of skin tone.

Google said in February that cameras on some Android phones could measure pulse rates via fingertip, and that the feature would work on light or dark skin.

The company has also made skin-type assurances for a feature for filtering backgrounds on Meet video conferences and an upcoming web tool called Derm Assist for identifying skin conditions.

Product testing

The tests for these products were conducted using the six-tone FST scale.

Google is ahead of competitors in developing its own alternative to FST.

Microsoft, Apple and Garmin, for instance, make use of FST to test health-related sensors in products such as smartwatches.

Microsoft told Reuters it was aware the system was imperfect, while Apple said it uses various measures as well as FST and Garmin said it believes the system is reliable.

Google is also planning to deploy an artificial intelligence feature this month that addresses the longstanding issue of darker skin tones being under-illuminated in video chats.