The research will focus on new ways to create and deliver object-based media at scale.

AI image 1

BBC R&D has partnered with the universities of Surrey and Lancaster to develop and trial new ways to create and deliver object-based media at scale.

Object-based media is content that is tailored for individual viewers by breaking it into separate ‘objects’ - which can be as large as the audio or video or as small as an individual frame - attaching meaning to them, and rearranging as fit using AI.

The eventual goal is to supply a wide range of content experiences to mainstream audiences through various platforms and devices.

This scheme is a Prosperity Partnership, a series of business-led research partnerships between leading UK-based businesses and long-term strategic university partners with matching funding from public body, UK Research and Innovation.

Prosperity Partnership aims:

 - Enabling real-time object extraction of audio-visual objects using AI, without having to capture content in a way specifically to support object-based media, which increases costs and can be impractical for many kinds of programmes

 - Enabling a distributed intelligent network compute architecture that will allow people to get an object-based experience without needing the latest consumer equipment and without needing massive cloud computing infrastructure.

The University of Surrey will be contributing its audio-visual AI expertise to develop AI-powered techniques to allow audio and video to be separated into objects (such as individual audio tracks and distinct ‘layers’ from a video scene) and create metadata describing the scene

Meanwhile, the Lancaster University will provide the software-defined network distributed computing expertise of its networking group to develop approaches to intelligently distribute the processing and data through the delivery network, making the best use of resources in the cloud, edge-compute nodes, and the audience’s networks and devices.