Where: Channel | Metadata | General tab
This feature is only available in Color Factory Enterprise
Color Factory can use the Azure Face service, part of the Cognitive Services on Azure, for face recognition. While the integration with the Face service is included in Color Factory Enterprise, you will need to subscribe to the service via the Azure portal and obtain an API key to use the service.
Subscribing to the Azure Face service
If you haven't already done so, learn how to subscribe to the Face API on the Azure Portal!
Setting up face recognition
Once you've subscribed to the service, create a Color Factory channel with the Metadata feature enabled. On the General tab in the Metadata feature, Access the Face feature by clicking on the Azure Computer Vision button.
The face detection feature is configured in the lower part of the dialog:
First off, enable face recognition services using the checkbox. Then fill in these required fields:
Azure key for Face API: Enter the API key obtain on the Azure portal here.
Unique ID string for person-group: When training the Face recognition service, Azure stores a set of faces in a group. If this is your first time using the service, choose a name and enter it here.
Region: Find the region in the list that corresponds to the region you chose when subscribing to the Face service in the Azure Portal.
Training the face recognition engine
Before you can start using the face recognition feature, you need to train it:
- Set the face recognition engine to Training mode by selecting the Training radio button seen in the screenshot above.
- Next, you will need to find pictures of the people you want the engine to recognize and tag these pictures with their name in a metadata field, for example using FotoStation. Generally, the better view of their face in the picture, the more secure the recognition will be.
- In the face recognition settings in Color Factory, choose the XMP field that contains the persons' names so that this information can be sent to the service along with the actual image data for training.
- Finally, run the tagged pictures through the Color Factory channel to commence training.
Reviewing the training results
Afterwards, you can see the list of persons stored in the Azure Face service by clicking on the Manage Persons button. That dialog will show you the persons you've trained it for and the number of pictures that form the basis of the recognition of each person. You can also remove individuals or everyone from the list.
Processing images with face detection
Having trained the engine, switch the operating mode to Detection using the radio button. Now, the XMP field dropdown list is used to set the field into which Color Factory will write the results of the face detection. We recommend storing names in bag fields, as this will make each name a separate searchable entry.
The confidence level can be set to automatic, or you can set a a fixed minimum confidence level manually by dragging the slider to set the level of confidence with which a person is recognized for the subject's name to be added to the metadata. There is also the option to add the confidence level itself to the metadata.
Remember to save the changes. Then start Color Factory service and send images through the channel.