MM5.4 AI Scene Configuration
- Stefan Fodor (Unlicensed)
- Gosia Creosteanu (Unlicensed)
MM5 supports AI Scene detection (by use of Azure Cognitive Services) and can be saved to an EditMultiComboValue metadata field.
Requirements and limitations
AI Scene detection can only be used for image assets.
It requires that the user has the role Ai_Add and visual feature Landmarks is enabled.
If the metadata is set to not be auto-translated, the translation needs to be also set up in the cognitive services.
Landmarks will not work if the image you're trying to have a landmark detected on, is a part of a collage.
This means that just adding one image of e.g. a dog next to an image of e.g. the Eiffel Tower, will make the landmark detection say that there are no landmarks in the image. This is a limitation of Azure Cognitive Services.
Configuration
Navigate to the MM5 Config Manager, expand the "AI Config" and define the following configuration:
Usage
Scene detection can be used both in the single and the multi-asset editor:
Example in the asset editor:
Example in the multi-asset editor:
