Curator supports AWS Rekognition Services and Azure AI Video Indexer Services to scan and analyse your media assets for faces and automatically tag the media asset's metadata with the person (personality) that is contained in the media asset (images and videos). This article will explain how to task an asset(s) to be scanned for automated metadata tagging.
It's assumed you have already built your custom library. If you haven’t, see the building an image library for custom face detection guide.
💡 Note
In order for Curator to recognise and detect faces in media assets ingested, a System Administrator must have installed AWS Rekognition Custom Face or Azure Custom Face Recognition.
How to scan assets and automatically tag with metadata
-
Right-click an asset(s) and select the relevant plugin from the menu.
-
Select ‘Scan file for custom faces’.
-
You will receive a notification to advise you the process has been requested.
Once complete, the metadata of your asset will be populated with a list of names in ‘FaceSearchResults’ (AWS) or 'Faces' (Azure).
💡 Note
By default, when sending a media asset for custom face detection, subclips will be created and face results will be shown on those newly created assets (as shown in the image above)
With Azure, the same scan also recognises celebrities from Azure's built-in celebrity library. Their names appear in 'Faces' alongside your custom faces.
How to scan assets for celebrities (AWS)
With AWS, celebrity recognition is a separate scan. It uses the built-in AWS celebrity library, so you do not need to build a custom library first.
- Right-click an asset(s) and select the relevant plugin from the menu.
- Select 'Scan File for Celebrities'.
- You will receive a notification to advise you the process has been requested.
Once complete, sub-clips will be created and the names of the people recognised will be shown in their 'Celebrities' metadata.
Note
A System Administrator must have installed AWS Rekognition.