Bankless Times
Ever AI facial recognition technology supports multiple use cases
HomeNewsEver AI facial recognition technology supports multiple use cases

Ever AI facial recognition technology supports multiple use cases

News Desk
News Desk
January 31st, 2023
Why trust us
Advertiser Disclosure

Consider Ever AI’s story one where a good idea at the right time became much more than originally imagined, CEO Doug Aley said.

Founded in 2013, Ever AI was initially envisioned as a provider of products that automatically organizes and secures photographs. It does that, to the tune of nearly 13 billion photos and videos for tens of millions of users across 95 countries, but it has grown to offer facial recognition technology with the most comprehensive real-life dataset of any private company in the world. Ever AI’s models have identified hundreds of millions of global clustered identities to date, with each one generated from mobile consumers. A patent pending technique of collecting tagged data and integrating it into models has produced what the company says is the largest tagged dataset ever used to build enterprise-ready face and object recognition APIs and mobile SDKscked by Khosla Ventures, Felicis Ventures, Icon Ventures, Cherubic Ventures, Transmedia Capital, and SV Angel.

“At the time, no one was using AI and machine learning to categorize photographs,” Mr. Aley said of the environment in 2013. “There was nothing commercially available.”

[caption id="attachment_118617" align="alignleft" width="240"]
Doug Aley[/caption]

After positive industry feedback, in 2017 Ever AI decided to  build out its facial recognition technology, adding to a suite of options covering many different uses cases. In retail environments, facial recognition allows stores to identify customers and offer them personalized promotions, simplify checkout with “pay by face” authentication, and even understand who makes up their customer base at different times of day. Ever AI combines facial and object recognition to provide retailers with an integrated online-to-offline solution that can combine product suggestions and automatic checkout. Faster shopping times and higher average order value are among the key benefits.

Financial institutions can integrate facial recognition as another authentication layer, or add it to their multi-factor authentication process for large transactions. Like in retail, the technology can also be used to identify age, gender and ethnicity to offer personalized product suggestions.

As system automation increases in prevalence, facial recognition technology can identify permissioned users of factory machinery and even vehicles. Additional applications are available for military, smart cities, identification management and mobile carriers.

As Ever AI’s consumer base grew, it helped establish the business side, Mr. Aley explained. As people were tagged in photos clusters formed which produced rich data.

In order to maximize its effectiveness the technology has to adapt to a number of factors, Mr. Aley said. Some facial recognition models are trained on Caucasian facial structures. Their accuracy can be lower when identifying people with African and Asian ancestry.

Then there is picture and video quality. Pixel density, lighting and focus are a few issues that facial recognition technology must contend with.

Mr. Aley said Ever AI’s models are constantly improved as more photos and videos are added to their data bases. The 12.64 billion contributions to date come from the following areas:

  • North America – 8.22 billion (65 per cent)
  • Latin America – 1.04 billion (8.2 per cent)
  • Europe/Middle East/Africa – 1.67 billion (13.2 per cent)
  • Asia Pacific – 1.71 billion – (13.5 per cent)

Chinese image sources have been hard to come by but as society becomes more liberalized they look forward to accessing larger data sources not only from China, but in other parts of the world as citizens of those regions gain wider access to popular technology, Mr. Aley said.

“The iPhone X has normalized facial recognition to the point companies are rushing to develop a product roadmap.”

Facial recognition is likely to remain a leading authentication factor because it is a completely passive one, requiring no effort from the user, Mr. Aley said. Combine that with location and other factors and you can deliver a more secure experience.

“The ideal from a consumer perspective is a completely frictionless transaction…” Mr. Aley said. “Where Starbucks recognizes me and takes my order.”

Contributors

News Desk
The latest news, comment and analysis from our crypto news desk.