Where Video Analytics is Concerned,
AI is the New Old Kid in Town
But is it really AI?
We’ve been talking about AI for donkeys, but there have been fundamental shifts of late that means it isn’t going to disappear again, ever. This decade’s big boys have realised there is money to be made from centralised AI solutions in your home hub through to driverless cars. It is seen as a money spinning frontier for companies hungry to flex their gold digging muscles.
In the past, AI was largely based around text data. But that ignores an estimated 70% of the world’s data which is unstructured video and audio. So sights are being firmly set on that video data and how to analyse and monetise it. This surely has to be good news for the owners of that very data.
That said, not all of this is strictly AI though. For me, it’s only really AI if it simulates intelligent behaviour or imitates human intelligent behaviour. What we are looking at (largely) today is video analysis and metadata generation with a little bit of machine learning tagged on the side. But it sounds cooler if you call it AI, right?
- Some really cool searching of video for people / things. The next two years will be full of things found in archives that we never thought we had.
- Some great metadata analysis stats that will come out.
- Limited “automatic editing” “AI”, limited machine learning of what people are interested to find and pre-processing results for that purpose.
- An explosion of company’s offering “nirvana” (but delivering somewhat less) in this arena.
The Literally Bad Pun:
Alec- Knock Knock jokes.”
This joke was reputedly written by an AI robot. Humans. We are safe for now.
Are trialling some really cool video search and metadata extraction technologies. View to launch during 2018.
Example AI aims from WIG:
Post-production house: When doing basic ingest, conform worklflows based on machine learning to understand what needs to be conformed .. frees up junior staff to work on other things.
Leading industry analyst: Web based agents aggregate current and trending news. Uses an API (like the one to MatrixStore) to seek out content in the archive that might be relevant. Builds a timeline based on the news .. presents editor with option to retrieve and edit. Imagine the time saved in building edits.
Some Useful Links:
About Object Matrix
Object Matrix provides digital content governance and object storage platforms. The company was built on the philosophy that archive systems should be scalable and interoperable, as well as ensuring instant access to data and metadata. Its flagship product, MatrixStore, is an integrated object storage software solution providing protection and governance for the lifetime of any digital content. It’s used by global organisations that create, curate, and distribute video content, including NBC Universal, TV Globo, the BBC and BT.
Object Matrix is comprised of a team of storage industry experts with a shared vision for high levels of data security combined with intuitive user interfaces. In addition to providing solutions and product integrations for the storage, search and retrieval of media assets, Object Matrix also provides consultancy, training, systems integration, and ongoing support services.