Automated Metadata Extraction – Where are we Now?

Posted on January 10, 2016

Pam Fisher


For many years, automated metadata generation for moving image content has been an attractive goal, but largely unrealized in modern systems. This talk combines a review of past initiatives, which help to illuminate requirements articulated over time, and discusses a long term project underway between BAFTA and University College London (UCL) – Video Clarity. This asks the question: “What can be learned by pointing a computer at video?” – what semantic, rights and engineering metadata can be discovered automatically? Video Clarity began as a ‘search and discovery’ technology project aiming to perform extremely rapid search of moving image files using embedded hidden metadata, and was soon able to search 100,000 files in under ten seconds. Two more things were needed: richer semantic meaning, and support for the modern metadata context of linked data and a multitude of standards. The presentation looks at the current state of metadata extraction, search, and open tools.