Turning on Facial Recognition


Please contact your Account Manager to enable this feature for your Canto account.


Once your account manager has enabled this feature for your account, please follow the below instructions to turn on the feature:


1) Go to Settings > Configuration Options > Recognition and select Facial Recognition.


2) Toggle the Facial Recognition option to On.

 


Images - Facial Recognition

Click Reidentify Faces (this is will identify any images that were already uploaded and any new images) and Save

 

With the Min. Confidence Score, you can control how accurate an image must match with your reference image before a face is accurately identified. 


You can set a limit on the maximum number of people to get recognized for each image. By default, it's set to 10 people per image but you can increase this value to up to 50.


There is also a Celebrity Recognition feature you may enable. 



Facial Recognition When Uploading Images

If Facial Recognition is enabled, Canto tries to recognize faces by default when uploading assets.

If you want to speed up the upload process, or if you do not need facial recognition for uploaded assets, you can deactivate facial recognition when uploading.


To turn off facial recognition when uploading:

  1. In the Add Metadata dialog that appears during the upload, scroll down to the Facial Recognition for Images section.
    You have two options:
  2. Untick the Use Facial Recognition... checkbox. This completely deactivates facial recognition when uploading.
    Or
  3. Just untick the Identify and tag new people checkbox. This only deactivates the (particularly time-consuming) recognition of new faces. Faces that are known to Canto, and that have already been recognized in other images, will nevertheless be recognized and indexed in this upload.



Video - Facial Recognition


Similar to Facial Recognition for Images, you can control the Confidence Score and control the maximum number of people to be recognized for each video. The length of the video and the number of people in the video will affect the time it takes for recognition software to process. 


For video recognition, you will have a preset number of hours of content you can identify. You can view the number of hours you have and the number of hours you have used. The number of hours can be adjusted after reaching to your account manager. Your hour usage is combined with the Smart Tag feature. 



When using the Facial Recognition feature for videos, you will need to manually identify video by selecting the desired video in the main library and click on the Reidentify icon. You will receive an email once this process is completed. 



Note: Facial Recognition for Videos does not have have Celeberity Recognition. Like any other person you are identifying, you will need to first identify and build an album of portraits. See below for best practices on how to do this.



Enabling Facial Recognition To Display


Once you are done with your desired configuration, navigate to Custom Configuration, and add the following:

  • Facial Recognition as a field under the information page

  • Facial Recognition on the metadata editor
  • People on the Filter Panel

And click Save.


Best Practices


Before you use the Facial Recognition feature, you will need to build a repository of portraits for your identified people by uploading an image that contains a full face and shoulders of the person. It is important to upload this first picture as a reference picture to help better identify pictures. Especially, for group photos. See below for sample images: