Amazon boosts Rekognition facial image capacity and performance

November 24, 2017 - 

Amazon has announced the addition of real-time face recognition across tens of millions of faces, and the capability of detecting up to 100 faces in “challenging crowded photos” to its Amazon Rekognition deep-learning image recognition platform ahead of its re:Invent conference, which takes place in Las Vegas next week.

The new capabilities also include text detection and recognition, according to the announcement, and Amazon says face verification and identification will improve by up to 10 percent for customers currently using Amazon Rekognition.

The real-time face recognition improvements amount to a reduction in search latency of five to ten times, and storage of 10 to 20 times more facial images.

“These improvements allow deputies in the field to receive the response to searches in near real time. This allows them to get the information they need and take action quickly. Seconds saved in the field can make the difference in saving a life.” Chris Adzima, Senior Information Systems Analyst for the Washington County Sheriff’s Office.
Face analysis and indexing from a crowded image increased from the previous maximum of 15, enabling accurate demographic data capture and sentiment analysis.

Amazon also announced plans to open ML Solutions Lab, a machine learning laboratory in which the company will collaborate with customers to develop new artificial-intelligence based technologies.

As previously reported, Novetta released a performance assessment of Amazon Rekognition in August.

Leave a Comment


About Chris Burt

Chris Burt is a writer and contributor to Biometric Update. He has also written nonfiction about information technology, dramatic arts, sports culture, and fantasy basketball, as well as fiction about a doomed astronaut. He lives in Toronto. You can follow him on Twitter @AFakeChrisBurt."