Aitheantas Aghaidhe

This article explains how to install, configure and use the Facial Recognition feature in visitor id.

Quick links:

• Introduction
• How Does It Work?
• What About Privacy?
• Important Notes
• System Requirements
• Installation
• Configuration

Introduction

The Facial Recognition feature in visitor id allows your Registered Visitors and Staff to sign-in and sign-out by recognising their face via a webcam in Kiosk mode. This feature was introduced in visitor id v7.0.

The Facial Recognition feature uses a free and open-source, real-time facial recognition service called Exadel CompreFace which is licensed under the Apache 2.0 license model.

The visitor id Facial Recognition feature use is limited to:

• Aged Care facilities
  • Staff
  • Contractors
  • Visitors, including family, friends and carers of the residents.
• All other facilities
  • Staff
  • Contractors

Any other use other than the intended limited use is a breach of the End User License Agreement.

How Does It Work?

The visitor id Facial Recognition feature is available in Kiosk mode only. The kiosk shows the webcam video at all times, waiting for a face to be found. The Kiosk takes a picture once every 2 seconds and checks if a face is in the image. If a face is found then the Kiosk will attempt to match the face in the image to the database of uploaded photos. If a match is found then the person can continue signing-in or signing-out. If the face is not recognised then a message is displayed to the person and an alert can be sent to certain email addresses and mobile phone numbers to advise that an unrecognised face attempted to sign-in.

What About Privacy?

We take data privacy seriously and have taken steps to ensure the database of photos is kept anonymous and password protected.

Photos of unrecognised faces are not stored in any database. The photos of unrecognised people are deleted straight away from the Kiosk PC.

The CompreFace database only stores photos of people that are to be matched for facial recognition. The photos do not have any personal identifying information stored with the, such as names, companies, email addresses, mobile phone numbers etc.

visitor id also contains functions to delete one or all of the photos from the CompreFace database.

It is the end user’s responsibility to ensure that the CompreFace software is behind a suitable firewall, not published to the intenet and only systems administrators have access. Failure to comply is considered a breach of the End User License Agreement.

Important Notes

The visitor id Facial Recognition feature can be used for either Staff, or Registered Visitors or both. The Barcode/QR Code Scanning feature will be disabled for Personnel, or Registered Visitors, or both entirely, depending on the enabled Facial Recognition features.

The photo used for the Facial Recognition match will also be used on the printed visitor pass (if using visitor passes).

When the Facial Recognition feature is enabled for Registered Visitors, new visitors that have not yet been pre-registered will not be able to sign-in or sign-out via the Kiosk.

System Requirements

• Docker and Docker Compose (or Docker Desktop).
• Exadel CompreFace. CompreFace could be run on most modern computers with x86 processor and AVX support. To check AVX support on Linux run lscpu | grep avx command.
• A supported Webcam.

Installation

1. Upgrade visitor id to v7.0.
2. Decide on which PC/Server the CompreFace software is to be installed on:
   • On visitor id standalone installations it is recommended to install CompreFace on the master visitor id PC.
   • On kwiklook integrated installations it is recommended to install CompreFace on the master kwiklook server.
3. Install Docker and Docker Compose.
   • On Windows it is recommended to install Docker Desktop.
4. Install CompreFace.
5. Open the CompreFace login URL, by default: http://localhost:8000/login
6. Click the ‘Sign Up’ link and create the administrator user with a name, email address and strong password.
7. Login to CompreFace with the created user.
8. Under the ‘Applications’ menu click ‘Create New’.
9. Name the application ‘visitor id’ and click ‘Create’.
10. Click on the newly created application ‘visitor id’.
11. Under the ‘Services’ menu click ‘Create New’.
12. Name the service’ Facial Recognition’, choose the ‘RECOGNITION’ type and click ‘Create’.
    • This will create an API Key for the Facial Recognition service that is to be used in visitor id to connect to the CompreFace API.

Configuration

1. Copy the CompreFace Facial Recognition service API Key created in the ‘Installation’ steps above.
2. Run visitor id.
3. Click ‘File > Option > General > Functionality’.
4. In the ‘Other Features’ group, tick the ‘Enable Facial Recogition’ checkbox. This will enable the ‘Facial Recognition’ tab.
   
5. Click the ‘Facial Recognition’ tab.
   
6. Enter the settings for the CompreFace API:
   1. API URL: This is the URL of the CompreFace URL, by default: http://PCNAME:8000/api/v1
   2. API Key: Paste the API Key here that was copied in step 1 above.
   3. Match Threshold: This is a percentage used to match faces. By default this is 98%. We recommend not going lower than 95%.
   4. Unrecognised Alert Email: Enter all email addresses that need to be alerted when an unrecognised person has attempted to sign-in. If the ‘Email Alerting’ option is enabled, an email alert will be sent to these email addresses.
   5. Include Visitor Photo in Alert Email: This will attach the photo that was taken of the unrecognised person to the email alert.
   6. Unrecognised Alert SMS: Enter all the mobile phone numbers that need to be alerted when an unrecognised person has attempted to sign-in. If the ‘SMS Alerting’ option is enabled, an SMS alert will be sent to these phone numbers.
      1. Please note: Photos can’t be attached to an SMS alert.
7. A test email or SMS can be sent to confirm the settings are correct by pressing the ‘Test Unrecognised Alert Email’ or ‘Test Unrecognised Alert SMS’ buttons.
8. Click the ‘Kiosk > People’ tab.
   
9. To enable the Facial Recognition feature for Personnel and Staff, tick the ‘Enable Personnel Facial Recognition’ checkbox.
10. To enable the Facial Recognition feature for Registered Visitors, tick the ‘Enable Registered Visitor Facial Recognition’ checkbox.
11. Click the ‘Kiosk > Message Prompts’ tab.
12. In the ‘Sign-In Messages 1’ tab, it is recommended to update the ‘Kiosk Welcome Message’ to advise your visitors that Facial Recognition is enabled.
    
13. In the ‘Sign-In/Out Messages’ tab, the ‘Facial Recognition Not Found Message’ can be changed. The default message shown to visitors who are not recognised is: “Oops! You are not recognised in our system. Please wait here and one of our staff members will be with you shortly.”
    
14. Click ‘Save’ to save the settings.