Multimodal AI models from startup TwelveLabs will become available within Amazon Bedrock, the two companies announced.
TwelveLabs’ models Marengo and Pegasus will help Amazon Web Services (AWS) users generate insights and search for video content. The models possess advanced video understanding capabilities and utilise natural language to seek out videos across AWS’ library.
The technology from TwelveLabs, a San Francisco-based start-up, is a video understanding platform that uses AI to extract and search for meaningful content within videos
For example, a sports network would be able to segment video libraries by certain plays within a match. The models have the intelligence to identify patterns which make a library of video content more approachable to a user.
“Video contains nearly 80 percent of the world’s data, yet most of it remains unsearchable and underutilised.” Jae Lee, Founder and Chief Executive Officer of TwelveLabs, says. “Our technology enables users to search across [Amazon Bedrock’s] entire content library–from videos collected 10 years ago to 10 minutes ago–to find the precise moment they’re looking for in a single second, and then interpret and analyse those moments.”
Amazon Bedrock is a service within AWS that offers a range of AI models to users through a single API. AWS is the first cloud provider to use TwelveLabs’ models.
The tech giant said the service offers data protection for users as they build and use generative AI models. Additionally, users can build models which search through large video collections or summarise content.
The models simultaneously comprehend video, audio, and text. From a TV perspective, broadcasters can more efficiently distribute content with relevant video ads integrated into it.
“Video understanding is revolutionising how industries like media and entertainment, sports, automotive, and education work with and discover content.” Samira Panah Bakhitar, General Manager of Media & Entertainment, Games, and Sports at AWS, explains. “Businesses will now be able to easily search, categorise, and extract insights from their vast video libraries, enabling new use cases and better user experiences that were previously impossible without significant technical expertise.”
The partnership was announced at this years’ NAB Show in Las Vegas.
