The rise of machine learning in TV and film production

We’re seeing a rise in the emerging technologies of artificial intelligence (AI) and machine learning, and it’s reaching far beyond the production space. Machine learning applications in editing and post-production projects are coming to the fore, and the BBC recently made use of the technology in its first experimental BBC 4 documentary, ‘Made by Machine - When AI Met the Archive’.

Vast archives of footage, images, marketing materials and post production work can be used for machine learning applications, and in turn drive creative automation.

The technology also offers solutions to internal organisational problems. With the cloud and machine learning algorithms, businesses can begin to structure and automate processes for all of their content, saving hours of manual review and data entry.

Video intelligence can identify key speech and individual faces. This enables you to instantly search an archive of past productions that mention a specific word or feature an individual’s face. If you possess a large bank of images, machine learning can detect individual objects and recognise text in image files, automatically tagging recognisable words and faces. For audio files, you can apply audio intelligence to identify key words and topics, allowing you to search easily by theme or even single words within a file.

From there, producers can start to explore potential applications. Take crew management, for example, the process for managing freelancers and contractors can be streamlined by providing one central place for previous hire vital information. By assigning an automatic level of data classification to documents such as tax forms, work history and hourly rates, machine learning can tag contractor profiles with keywords, allowing teams to quickly find the information they need for a fast hire. As most projects depend on freelance support to deliver the final cut, building out this process can prove invaluable.

A similar mechanism can be applied when managing talent. Contract management is time-consuming, particularly when documents are in different formats and being handled by countless individuals – whether agents or internal staff. By creating instructions that only pull out the data you need, producers can automatically gather any required information from a scanned copy, photo or PDF of a contract; even the signature and renewal dates can be pulled automatically.

Finally, you have applications involving the film content itself. Within a production house’s archive there will be thousands of images, posters, promotional materials and tons of footage. All of these need to be catalogued, which is where machine learning comes in.

Production agencies host a treasure trove of unused digital content and machine learning drives efficiency using the intelligence and information that is already at an organisation’s disposal. To keep up with the growing demand for content from global audiences, content creators need to consider modernising and digitising their business processes. With the likes of IBM, Microsoft and Google integrating with tools like Box Skills, the solutions are there.

Jade McQueen is senior managing director, media & entertainment at Box. The company’s customers include Endemol Shine and Fremantle Media