Artificial Intelligence & Machine Learning

White Paper: An evaluation of generated/manipulated image detection

Marc Górriz Blanch, Woody Bayliss, Danijela Horak

Detecting manipulated media is crucial to combat misinformation.

White paper published May 2024

NeRFs - A guide to creating impressive 3D fly-through videos from 2D smartphone images

Creating complicated moving shots for small radio studios can be hard, so we tried using Neural Radiance Fields for some striking visual effects.

Blog post on 18 Jan 2024

Responsible innovation and society

Data and technology affect all aspects of the ��tv, from our editorial operations and outputs to how audiences can discover and access our services and content. The ��tv needs to...

Project

5 things we learnt about the barriers we face and bridges we must build to ensure Responsible AI

Launching the Bridging Responsible AI Divides programme with our partners.

Blog post on 26 Oct 2023

Addressing the cold start problem in recommender systems

��tv iPlayer has a constant stream of new users and content. How can we make recommendations when there isn't enough information about either?

Blog post on 14 Jul 2023

Parameter-reduced frame interpolation using machine learning

Motion blur can be caused by interpolating frames, algorithms can help with this problem.

Blog post on 5 Jun 2023

We need to talk about recommendations: helping curators and data scientists work better together

Communication is key to enabling curators and data scientists to work towards the same goals - but often recommendation systems are understood in very different ways.

Blog post on 17 May 2023

Is this real? Provenance in generative media

Knowing about the authenticity of the media we consume online means knowing where it came from and what has happened after it was created.

Blog post on 19 Apr 2023

AI-assisted journalism: our open-source quote extraction system

We are releasing an open-source quote extraction and attribution system called Citron.

Blog post on 19 Apr 2023

Data Science Research Partnership

The Data Science Research Partnership aims to put the ��tv at the forefront of machine learning and data science for the media industry, unlocking the potential of the ��tv’s data...

Project

Wing Watch – an AI enhanced interactive wildlife stream

Using AI-generated data to help the audience explore the wildlife streams from Winterwatch.

Blog post on 27 Jan 2023

Pruning complex neural networks to achieve faster video coding

Neural networks used in machine learning tools need many resources. Pruning removes network redundancies to make tools more efficient and accessible.

Blog post on 25 Oct 2022

Improving pixel quality of compressed video

A promising new way of compressing video, using innovative neural networks.

Blog post on 24 Aug 2022

White Paper WHP 402: Old School: An 8K Multicamera Shoot to Create a Dataset for Computational Cinematography

Graeme Phillipson, Stephen Jolly, Alia Sheikh, Peter Mills, Rebecca Stagg, Michael Evans

R&D designed an 8K multi-camera production, to create a dataset to help further research into AI automated editing systems

White paper published Aug 2022

Automatically re-lighting scenes in post-production - using machine learning

We're investigating whether it's possible in post-production to automate the re-lighting of footage for events that don't have a dedicated lighting crew.

Blog post on 16 Jun 2022

��tv

Accessibility links

Research & Development