TwelveLabs video understanding models are now generally available on Amazon Bedrock. This integration brings advanced AI capabilities for video analysis directly to AWS users. The new offering includes Marengo for video embeddings and Pegasus for video language generation.
Marengo, a video embedding model, excels at search and classification tasks. Pegasus, a video language model, generates text from video data. Both models leverage training on Amazon SageMaker HyperPod. This delivers groundbreaking video analysis with precision and reliability.
Users can now perform natural language video searches, instantly jumping to specific moments like "show me the first touchdown of the game." The models also generate descriptive text such as titles, topics, hashtags, summaries, chapters, or highlights. This enables deeper insight discovery from vast video libraries without requiring predefined labels. Businesses can, for example, identify recurring themes in customer feedback or spot subtle product usage patterns that were previously unobservable. This transformation turns entire video libraries, whether hundreds or thousands of hours, into a searchable knowledge resource. It maintains enterprise-grade security and performance, crucial for large-scale deployments.
Empowering Video Workflows with TwelveLabs on Bedrock
Industries across the board benefit from these new capabilities. Media producers and editors can instantly locate specific scenes or dialogue, allowing them to focus on creative storytelling rather than sifting through hours of footage. Marketing teams streamline their advertising workflows by quickly personalizing content to resonate with various audiences. Security teams proactively identify potential risks by spotting patterns across multiple video feeds, enhancing situational awareness and response.
Integration is straightforward via the Converse API within Amazon Bedrock. Developers can leverage the AWS SDK for Python (Boto3) for seamless application building. Marengo Embed 2.7 and Pegasus 1.2 models are now accessible.
