AWS for M&E Blog

Media intelligence just got smarter with Media2Cloud 3.0

We are happy to announce the official release of Media2Cloud 3.0. This release helps AWS customers simplify and expedite their media migration workflow into AWS. It provides a framework to implement artificial intelligence and machine learning (AI/ML) to create frame-level descriptive metadata about content. This new version of Media2Cloud still includes the features that customers have used since its first release in 2018. For example, features like serverless checksum validation, enhanced technical metadata generation, automated video, image and audio proxy generation, and granular metadata extraction with Amazon Rekognition and Amazon Transcribe.

AWS Customers with digital media assets can use the Media2Cloud framework to create an automated content processing pipeline. When files are uploaded to an Amazon Simple Storage Solution (Amazon S3) bucket, the Media2Cloud workflow automates key steps like creating standardized metadata, identifiers, proxies, and adding machine learning metadata to content as a foundation. This makes it easier to manage and search for all assets as libraries or archives grow.

AWS Partners can accelerate the migration of content to AWS for customers by leveraging the Media2Cloud framework. AWS Partners are certified to deploy and customize Media2Cloud based on a customer’s requirements. This can include taking the Media2Cloud output and integrating AI generated metadata with Media Asset Management platforms. AWS Partners can also create custom AI/ML models for use by Amazon Rekognition Custom Labels. We encourage any AWS Partner interested in implementing or customizing Media2Cloud for their customer implementation to reach out to the AWS Partner team about how they can become certified.

Media2Cloud history

Media2Cloud is an AWS Solution designed to provide a structured process for getting video, image, and audio content under management within AWS. It’s a serverless ingest and analysis framework that considers the common attributes of on-premises ingest workflows to ensure new video assets are processed and supported with consistent metadata and proxies. The framework provides customers a way to avoid weeks of setup and configuration, and provides a starting point where customers and partners can modify the framework to meet their business objectives.

Media2Cloud covers the standard essentials for ingesting video, image, and audio content like assigning a UUID, running a MD5 checksum, technical metadata extraction, and the creation of proxies of thumbnails. In addition to this process, the framework includes a trigger to augment the baseline metadata of the video, audio, and image assets with AWS Machine Learning. The asset will have object and face recognition performed with Amazon Rekognition, speech to text is created via Amazon Transcribe, contextual metadata is created using Amazon Comprehend, and printed text and handwriting can be extracted from a document via Amazon Textract. The Media2Cloud web user interface provides users with a way to search for and preview content based on the generated metadata. The service is elastic, meaning the same workflow can be used to support day-to-day production and archive migration ingest as long as the requirements are the same. There’s no need to create separate workflows to accommodate capacity, a common issue with on-premises solutions.

What’s new

We continue to listen to our customers and are working to enhance Media2Cloud features and ease of use. In conjunction with AWS Professional Services, Formula 1 (F1) used Media2Cloud to migrate more than 4 petabytes of archived video content to Amazon S3. F1 continues to use the Media2Cloud framework to bring content from the current season into Amazon S3 and uses AI/ML services to reduce the time it takes to catalog content. You can learn more about F1’s use of Media2Cloud in AWS Solutions Media2Cloud: Architecture for the Formula 1 Legacy.

Version 3.0 of Media2Cloud now includes:

  • An enhanced Advanced Search that displays detection type and a timestamp so that a user can jump to that specific point in time in the video asset.

The New Advanced Search window of the Media2Cloud user interface.

  • A Stats page to provide a user with the overall categorization and top detections found in a content library.

A view of the Aggregated AI/ML detections on the Media2Cloud Stats page.

  • The Integration of the Amazon Rekognition Video Segment API. This allows a user to analyze video to detect Shot Changes and Technical Cues such as black frames, Society of Media Professionals, Technologists and Engineers (SMPTE) color bar, and end credits. Version 3.0 also converts segment results into an Edit Decision List (EDL) format, allowing you to import the shot timeline to popular editing software such as Adobe Premiere Pro and BlackMagic DaVinci Resolve.
  • The Integration of Amazon Rekognition Custom Labels (CL), an automated ML service that lets you easily train computer vision models such as image classification and object detection model. Version 3.0 can use your trained CL model to analyze content. Version 3.0 also manages the runtime of your CL models (auto-start and stop the model after use) to minimize inference cost.
  • Media2Cloud Version 3.0 users can activate Frame Based Analysis on the AI/ML settings page. The Frame based analysis option lets you specify the frame rate (i.e., one frame every two seconds) to use the Amazon Rekognition Image API instead of the Amazon Rekognition Video API.
  • The integration of Amazon Transcribe Custom Language Model (CLM) and Amazon Transcribe Custom Vocabulary. This gives a user the ability to bring their own CLM and provide accurate spelling of complex words or phrases to improve transcription results.
  • Media2Cloud 3.0 can auto-start the ingest workflow when a file is uploaded to Amazon S3. This allows the use of other transfer services to move files to the S3 bucket.
  • Media2Cloud 3.0 implements a Backlog Management System to support large requests of the Amazon Rekognition Video API analysis, which allows users to batch process archive video files without worrying about service quotas.
  • Media2Cloud 3.0 helps users to index faces for identification. When playing video, a user can enter snapshot mode and draw a bounding box around a face. Then, the user names the face and adds it to the Face Collection. The faces in a Face Collection can be indexed and media can be reanalyzed to gather additional detections of that person.

These new features will continue to help organizations dealing with media content accelerate their journey to AWS. The additional AI/ML services provide richer metadata and depth for users looking for a specific piece of content.

Participating AWS Partners

The Media2Cloud framework continues to grow in the partner space. With over a dozen AWS Partners certified, we are pleased to add two additional certified partners, Quantiphi and Signiant.

Quantiphi

Quantiphi is an AWS Partner with deep experience in AI-first digital engineering solutions. These solutions have helped media and entertainment customers unlock hidden data potential to curate better content, enhance customer targeting, and implement effective channel strategies. Each of Quantiphi’s category defining solutions are designed to transform the customer experience. Using Media2Cloud, Quantiphi has developed a people-pathing solution that ingests and stores content on AWS and uses Amazon Rekognition APIs to identify persons of interest and their associated metadata in video assets.

Signiant

Signiant is an AWS Partner that offers intelligent file transfer software trusted by more than 50,000 companies to move petabytes of high value content every day with speed, reliability and security. No matter where content needs to flow — between people, between systems, with partners, or to and from the cloud — Signiant has it covered. Each of Signiant’s products leverages proprietary acceleration technology, moving content up to 100x faster than standard Internet transmission speeds. Signiant’s acceleration technology is capable of moving any size file or data set over any IP network, while taking advantage of all available bandwidth. Signiant customers using MediaShuttle or Signiant Jet can deliver media files to Amazon S3 storage. Once the files transfer, the Media2Cloud workflow begins automatically.

Conclusion

Media2Cloud Version 3.0 is available on the AWS Solutions page. Follow the Implementation Guide to create the workflow on your AWS account. For more information about AWS Partners that help you deploy and customize Media2Cloud, check out the Media2Cloud Solution page

Jason Dvorkin

Jason Dvorkin

Jason is a Senior Industry Specialist BD supporting customers in the Media & Entertainment Vertical for AWS. Jason works with executives and technology owners to address their technical and non-technical transformation challenges, providing domain expertise for cloud adoption and achievement of key business objectives. Jason started his career producing live events and broadcasts for a professional sports team. He then went on to work for multiple tech start-ups in the areas of media asset management and media workflows orchestrion and machine learning and artificial intelligence. When Jason isn’t delivering results customers, he can be found spending time with his wife and two daughters, or on the ski slopes in Colorado.

Aramide Kehinde

Aramide Kehinde

Aramide Kehinde is a Sr. Partner Solution Architect at AWS in Machine Learning and AI. She is Lead for AWS Media Intelligence, which is an offering of solutions that help customers integrate their content workflows with AIML capabilities. Her career journey has spanned the areas of Business Intelligence and Advanced Analytics across multiple industries. Outside of work, she co-leads a Chicago non-profit with the mission of empowering creatives and technologists to create and build more. She also enjoys exploring the intersection of storytelling and AI.