Tag Archives: chronological age with the age predicted by a deep neural network

Robots judge a beauty contest

I have a lot of respect for good PR gimmicks and a beauty contest judged by robots (or more accurately, artificial intelligence) is a provocative idea wrapped up in a good public relations (PR) gimmick. A July 12, 2016 In Silico Medicine press release on EurekAlert reveals more,

Beauty.AI 2.0, a platform,” a platform, where human beauty is evaluated by a jury of robots and algorithm developers compete on novel applications of machine intelligence to perception is supported by Ernst and Young.

“We were very impressed by E&Y’s recent advertising campaign with a robot hand holding a beautiful butterfly and a slogan “How human is your algorithm?” and immediately invited them to participate. This slogan captures the very essence of our contest, which is constantly exploring new ideas in machine perception of humans”, said Anastasia Georgievskaya, Managing Scientist at Youth Laboratories, the organizer of Beauty.AI.

Beauty.AI contest is supported by the many innovative companies from the US, Europe, and Asia with some of the top cosmetics companies participating in collaborative research projects. Imagene Labs, one of the leaders in linking facial and biological information from Singapore operating across Asia, is a gold sponsor and research partner of the contest.

There are many approaches to evaluating human beauty. Features like symmetry, pigmentation, pimples, wrinkles may play a role and similarity to actors, models and celebrities may be used in the calculation of the overall score. However, other innovative approaches have been proposed. A robot developed by Insilico Medicine compares the chronological age with the age predicted by a deep neural network. Another team is training an artificially-intelligent system to identify features that contribute to the popularity of the people on dating sites.

“We look forward to collaborating with the Youth Laboratories team to create new AI algorithms. These will eventually allow consumers to objectively evaluate how well their wellness interventions – such as diet, exercise, skincare and supplements – are working. Based on the results they can then fine tune their approach to further improve their well-being and age better”, said Jia-Yi Har, Vice President of Imagene Labs.

The contest is open to anyone with a modern smartphone running either Android or iOS operating system, and Beauty.AI 2.0 app can be downloaded for free from either Google or Apple markets. Programmers and companies can participate by submitting their algorithm to the organizers through the Beauty.AI website.

“The beauty of Beauty.AI pageants is that algorithms are much more impartial than humans, and we are trying to prevent any racial bias and run the contest in multiple age categories. Most of the popular beauty contests discriminate by age, gender, marital status, body weight and race. Algorithms are much less partial”, said Alex Shevtsov, CEO of Youth Laboratories.

Very interesting take on beauty and bias. I wonder if they’re building change into their algorithms. After all, standards for beauty don’t remain static, they change over time.

Unfortunately, that question isn’t asked in Wency Leung’s July 4, 2016 article on the robot beauty contest for the Globe and Mail but she does provides more details about the contest and insight into the world of international cosmetics companies and their use of technology,

Teaching computers about aesthetics involves designing sophisticated algorithms to recognize and measure features like wrinkles, face proportions, blemishes and skin colour. And the beauty industry is rapidly embracing these high-tech tools to respond to consumers’ demand for products that suit their individual tastes and attributes.

Companies like Sephora and Avon, for instance, are using face simulation technology to provide apps that allow customers to virtually try on and shop for lipsticks and eye shadows using their mobile devices. Skincare producers are using similar technologies to track and predict the effects of serums and creams on various skin types. And brands like L’Oréal’s Lancôme are using facial analysis to read consumers’ skin tones to create personalized foundations.

“The more we’re able to use these tools like augmented reality [and] artificial intelligence to provide new consumer experiences, the more we can move to customizing and personalizing products for every consumer around the world, no matter what their skin tone is, no matter where they live, no matter who they are,” says Guive Balooch, global vice-president of L’Oréal’s technology incubator.

Balooch was tasked with starting up the company’s tech research hub four years ago, with a mandate to predict and invent solutions to how consumers would choose and use products in the future. Among its innovations, his team has come up with the Makeup Genius app, a virtual mirror that allows customers to try on products on a mobile screen, and a device called My UV Patch, a sticker sensor that users wear on their skin, which informs them through an app how much UV exposure they get.

These tools may seem easy enough to use, but their simplicity belies the work that goes on behind the scenes. To create the Makeup Genius app, for example, Balooch says the developers sought expertise from the animation industry to enable users to see themselves move onscreen in real time. The developers also brought in hundreds of consumers with different skin tones to test real products in the lab, and they tested the app on some 100,000 images in more than 40 lighting conditions, to ensure the colours of makeup products appeared the same in real life as they did onscreen, Balooch says.

The article is well worth reading in its entirety.

For the seriously curious, you can find Beauty AI here, In Silico Medicine here, and Imagene Labs here. I cannot find a website for Youth Laboratories featuring Anastasia Georgievskaya.

I last wrote about In Silico Medicine in a May 31, 2016 post about deep learning, wrinkles, and aging.