google: Google Photos get Real Tone Filters: What are they and how they work – Times of India

NEW DELHI: Google announced Real Tone filters at 2022 edition of its annual developers conference, I/O 2022. These filters are designed to work across skin tones. In 2021, Google announced Real Tone for Pixel, this year it has expanded them to more Google products. “Today, we’re introducing a next step in our commitment to image equity and improving representation across our products,” said the company while making the announcement at I/O.
Google announced Real Tone filters in partnership with Harvard professor and sociologist Dr Ellis Monk. The company has made the Monk Skin Tone (MST) scale, a 10-shade scale that the company says is culmination of Dr Monk’s research into the area. Named after Dr Monk, the MST Scale is composed of 10 shades, a number that the company says is chosen so as not to be too limiting, but also not too complex. These skin tones can be applied to machine learning models for computer vision projects of various kinds.
What are these Real Tone filters
Real Tone filters coming to Google Photos means that users can choose from a wide variety of looks/skin colour tones and pick the one that reflects their style. These filters are coming to Google Photos across Android, Apple iOS and Web. This means that now these filters will work even on photos taken on devices other than Google’s Pixel phones, including iPhones. These Real Tone filters can also adjust photos that users have taken previously to better reflect their skin tone.

Beyond Google Photos, the company is also rolling out enhanced support for skin tones through Google Search. When a user searches for images, such as “bridal makeup looks,” Google Search will automatically show a handful of tone options on applicable searches.
How is Google using MST scale
Google says that it is using the MST scale to evaluate how well ML (machine learning) models perform at detecting people’s faces within images, across skin tones. The company also says that it has developed some best practices along the way. Teams within Google are said to be incorporating skin tone fairness analysis throughout the product development cycle, including for ML modeling.