Edge Video AI

Sceney / Ads

Ad Enrichment Feed

Watches for ad breaks in the live stream and shows what's on screen going into each one, as IAB categories, scene keywords and brands seen on screen.

It follows the programme

Sceney reads the live stream continuously: the picture, what is said, and the text and logos on screen.

It knows when the break comes

When the stream signals an ad break, the feed holds the programme context from just before it. The adverts never overwrite it.

It shows what was on screen

Categories, keywords and brands for the moment going into the break, ready for the ad system to use.

What one read carries

IAB categoriesFrom the IAB Tech Lab Content Taxonomy, in the version your ad system asks for.
Scene keywordsShort phrases for what is happening, grounded in the picture and in what is said.
Brands seen on screenRecognised from logos, boards and products in the picture.
Break stateWhether a break is in progress, so context is used at the right moment.

Designed to be safe to switch on

Fails open

If the context is stale or empty, the answer is empty and the ad request carries on exactly as before.

Standard vocabulary

Categories come only from an allowlisted taxonomy. Answers that do not match are discarded, not repaired.

Programme, not person

The feed describes what is on screen. It does not identify or profile viewers.

Measured, with scope

What we measured is in the report.

Technical report 02 describes the live enrichment pipeline and the internal measurements we can publish, each with its scope and limits. A labelled accuracy study is still future work. Details are in Technical report 02.

See it on a live stream

The Ad Enrichment Feed runs on Sceney.

Try Sceney