Views: 236
Number of votes: 5
Average rating:

New add-on in the nuget feed: Enrich your image assets with Azure Cognitive Services.

Back in November as Ascend Nordic, I demoed an early version of this functionality. It was later improved and shown at the opening keynote of Ascend USA. I received quite a lot of requests after that to make it publicly available - and now (I admit a bit late, Jeroen even beat me to it) I'm now sharing it as a nuget package on the Episerver feed - as well as putting my code on GitHub.

Basically, this is how it works:

1) Install the nuget package on your Episerver site. (Episerver.Labs.Cognitive). Note that this is an experimental package and not supported in any way!

2) Add the access key to your appSettings in web.config (details are included in a readme).

And start adding the properties you want to your image model and attach attributes to them to describe what kind of cognitive content you want them populated with!

For instance, to add a descriptive text to your image (can be useful for indexing with Episerver Find or Vulcan) just add this attribute:

[Vision(VisionType =VisionTypes.Description)]
public virtual string Description { get; set; }

To index text recognized in the image add:

[Vision(VisionType=VisionTypes.Text)]
public virtual string TextRecognized { get; set; }

Or, one of my favorite features - Get a smart thumbnail (where the interesting parts of the image are in focus):

[ScaffoldColumn(false)]
[SmartThumbnail(100,100)]
public virtual Blob SmartThumbnail{ get; set; }

You can of course also extract a list of the age and gender of all people in the image like this:

[Vision(VisionType = VisionTypes.Faces)]
[BackingType(typeof(PropertyStringList))]
[Display(Order = 305)]
[UIHint(Global.SiteUIHints.Strings)]
public virtual string[] Faces { get; set; }

and much, much more.

Enjoy!

  Please login to post a comment