Amazon Web Services (AWS), recently upgraded its Amazon Rekognition AI solution. This increased its ability to recognize faces in crowded photos and added a text-identification feature.
Rekognition, an image-recognition tool, was launched at the 2016 AWS reInvent conference. It allows users to search for and index real-world objects, faces, and scenes in photos. It can also detect facial attributes such as gender, expression, and distinguishing facial characteristics.
AWS announced last week an upgrade to facial-recognition capabilities. Rekognition is now able to search for specific faces against millions of photos in real-time. AWS announced that this would result in a 5-10X decrease in search latency and allow for collections that can store between 10-20X and 20X more faces than ever before.
AWS pointed out that law enforcement agencies can benefit from the real-time search capability. They are always looking for ways to speed up the process of identifying suspects.
Rekognition can now identify up to 100 faces in busy or crowded photos, as opposed to the previous limit of 15. This capability can be used by businesses like department stores to gain insight into customer sentiment and demographics.
AWS now has the ability to detect text (such street signs or license plates) in real-world images. This includes photos, stills, TV reports, and social media posts.
The new “Text in Image” feature allows users to search and catalog images that contain the same text. It supports text in most Latin scripts and numbers embedded into a wide variety of layouts.
