Media Intelligence Explained - Maarten Verwaest of Limecraft in conversation with Stan Moote of IABM

In this interview, Maarten Verwaest, co-founder at Limecraft (www.limecraft.com) and Stan Moote, CTO at IABM (theiabm.org/) discuss the concept of Media Intelligence - blending AI with the right human input, and how a good implementation of AI contributes to a reduction of carbon emissions and improving your environmental sustainability goals. Full backstory on the blog of Limecraft:
www.limecraft....
00:00:15 What is Media intelligence?
Limecraft has been working with artificial intelligence since long. Our customers voiced frustration around how AI is implemented in their day to day business. Results are often inaccurate and incomplete.
Media intelligence takes into account any pre-existing data and knowhow available in the production process to get a more accurate result.
00:01:42 - How do humans fit in the loop?
Media Intelligence is not just about feeding the engine with any available documents coming from pre-production. We also take into account the intentions of the journalist or documentary maker. This is an iterative process. The documentary maker, the transcriptionist, and the journalist constantly feed the system. This has nothing to do with machine learning as such.
00:03:42 - How did you come to this idea?
It started by doing fundamental research for VRT R&D or VRT Medialab back in 2008. Limecraft didn't exist yet. We assumed that professional producers, working within tight constraints. Fiction or TV series is based on a scenario, a screenplay. Often, there will be no more than 5 or 6 protagonist players in the images. If you take this into account and you feed those constraints into the engine, the chances that it returns an accurate description are much higher.
So Limecraft was not a lucky shot. Media Intelligence is the result of what we've been working on for the last 20 years.
00:07:24 - How is Media Intelligence related to reducing carbon emissions?
From the very beginning, Limecraft's commercial model was based on the number of hours of video managed and processed and not per CPU or per user, unlike some BigTech companies.
By doing so, Limecraft created a mutual interest in minimising the operational footprint. We strive to minimise the number of CPU cycles needed to transcode, to transcribe, and to process in general.
So Limecraft is leading by example. We wanted to put a stake in the ground and be a good example of how we can help the entire industry optimising the environmental impact of the industry as a whole.