Articles on: Functionalities

Use Face Recognition [AI]

The application of the developed technology allows for easy time tracking for businesses and convenient, non-intrusive registration for employees. The technology ensures software security and ease of use, solving the issue of unauthorized time registration. By utilizing artificial intelligence methods, inEwi recognizes the face of the user assigned to the QR card.


Initial Configuration

In Settings (account icon in the top right corner of the web application), go to the Time tracking section. Enable Use Face Recognition.



‼️ Note that with the facial recognition feature enabled, the offline mode for the Time Clock by Unrubble application is not available. ‼️



Adding a New Face

During the first card scan by an employee after activating the feature, the system will prompt for face scanning.


Face registration can be done on mobile devices based on Android and iOS systems as well as in Kiosk mode through a browser on a computer.


Steps for face registration:

  1. The employee approaches the device so that the face and QR card are visible.



  1. Upon recognizing the QR card, the system will start saving the face. The employee should position themselves so that the face is fully visible within the circle.



  1. The system will display an appropriate message and complete the face registration.


  1. That's it! The employee can now register their work time, and inEwi will take care of the rest.



Employee Event Registration

  1. The employee with the QR card approaches the device serving as a reader (tablet/smartphone/PC).
  2. They show the QR card along with their face to the camera.
  3. The system records the event or displays an appropriate message.


Learning Process

During the initial employee scans, especially when done in different locations, the system may require confirmation (learning) of employees' faces.
You will see all events that require such confirmation in the Time Tracking > Clock Log.


During the first few days of scanning, the AI model learns the face of each employee and shows a special icon:



The learning process may take about 3 weeks on average, assuming standard 2 scans per day from Monday to Friday.
During this time, other employees may still scan in, even if they don't belong to that card.
Incomplete events remain with the learning process icon forever, indicating that this specific event went through the learning model.
Once the face model is fully trained, the icon for subsequent events will change to Verified. With such a model, there is no longer the possibility of another employee scanning in:



If the system detects abuse, such as an attempt to scan in with a photo instead of a face, the event will be rejected or wait for manual verification (accept/reject).
All rejected events also have their icon:




Fraud Detection

We divided this feature into two options:


Use Basic Fraud Detection

Our artificial intelligence will check if there is a real face on the other side of the camera.


Use Advanced Fraud Detection

The functionality from the basic version, additionally, our artificial intelligence will check the vitality of the employee's face.
The advanced option depends on the basic one.


‼️ Also, remember to update the Time Clock by Unrubble application on your device ‼️



General Issues

Technical Requirements

The face recognition AI system does not require additional devices or expose to additional costs.
Mobile devices such as tablets/smartphones handle this functionality well, even with a camera resolution of at least 480x360px.


Android

  • Minimum requirements: quad-core processor, 1 GB RAM, Android 5.0 or newer, front camera min 0.3 mpix
  • Recommended requirements: quad-core or octa-core processor, 2 GB RAM, front camera min 2.0 mpix, Android 7.0 or newer – the newer, the better 👌


⚠ An Android-based device must have access to Google services (Play Store) to download the application and its subsequent updates.


iOS

  • For iOS, the application performs well on any Apple device running iOS 15 and above.


Kiosk Mode (PC Browser)

We recommend using the latest versions of browsers such as:

  • Google Chrome - recommended
  • Opera
  • Mozilla Firefox
  • Microsoft Edge
  • Safari

Updated on: 28/06/2024

Was this article helpful?

Share your feedback

Cancel

Thank you!