Amazon will reportedly release underlying tech for real-time video recognition from its Amazon Go project and make it available to the public through its AWS platform.
According to VentureBeat, the technology won’t be unlike the Google’s recently released Cloud Video Intelligence API, which can recognize objects within video frames. But Amazon’s video recognition application will work for live IP-based broadcasts.
Amazon’s Go technology, which could see release on AWS later this year, will reportedly allow users to search for colors, objects and specific bits of audio within videos. AWS could end up integrating the new technology with Kinesis Streams, which already allows AWS users to search through streaming data.
Amazon Go is the company’s current trial involving stores without checkouts. Customers with the Amazon Go app enabled are able to walk into the stores, take what they want and walk out, while Amazon’s recognition engine automatically sees what they’ve taken and records the purchase.
Amazon has continued to build out the video analytics and production capabilities within AWS. In March, AWS acquired ThinkBox, a video and graphics processing firm used by media and entertainment companies.
Before that, in 2015, AWS paid $500 million to buy Elemental Technologies, a firm specializing in software for formatting OTT video.
"Elemental's deep media processing expertise at [the] high end of the market combined with AWS cloud will allow customers to focus the majority of their time on innovations," said Kalyanaraman Prasad, VP Edge Services at AWS, on a conference call with media following the announcement. "Our highly complementary solutions eliminate the cost and complexity that media companies have had to deal with in the past and leave the heavy lifting of the underlying technology structure to AWS."