Automated Facial Recognition

eMediaWorkspaces integrates Face Recognition on Images and Video assets making use of leading Face Recognition Open Source System: Exadel CompreFace.

We use CompreFace REST API to properly identify and group Faces we found in assets. Recommended installatio is CopreFace Docker image.

CompreFace Installation

  • Download latest CompreFace realease in a folder:
  • Exctract ZIP content and cd into exactracted path
  • Make any required customizations in file: docker-compose.yml
    • ​Recomended port configuration in compreface-fe service is: 8100:80
  • Edit .env file for more configurations
    • Recomended Max file size is: max_file_size=10MB
  • Run command: docker-compose up -d
  • Configure nginx to serve in the port 8100

CompreFace Configuration

Access CompreFace in browser and create and account.

Now create a Project and Copy the API Key the system generates.

In your eMediaWorkspace instance go to Catalog Settings (Settings | Preferences | Catalog Settings) and update or create the values:  faceprofileserver  and  faceapikey with your CompreFace server and API key values.

Enable Face Scan Event in your eMediaWorkspace Scheduler (Settings | Scheduler)

You can see Face Recognition results as they go in the Face Profiles Entity Group

You can customize some detecting Face configurations, add the desired configuration as variable in your Catalog Settings with your custom values:

Setting Default Value Description
facedetect_profile_confidence 0.99 similarity that on that image predicted person
facedetect_detect_confidence 0.999 minimum required confidence that a recognized face is actually a face. Value is between 0.0 and 1.0.