This guide contains updates related to Minor release April 2020.
Facial recognition is a way of identifying key features on a person's face and using it as a way to identify and differentiate between people. In WISEflow it is possible to use facial recognition as a way to identify if the participants sitting through an exam are really who they say they are, by using pictures of the participant from previous exams to cross-examine faces to prove, the participant has been the same person for every exam.
Furthermore, besides the initial authentication via facial recognition managers will be able to monitor the participant through snapshots of webcam recordings that will analyse the person’s identity throughout the exam. The actual snapshot and matching will be hidden from the participant but will result in a match score showing in the FLOWmonitor.
On the flow page for students, it will be noted for the students, that they will need a webcam for the exam.
When you first have the FLOWRecognition module on your license, it will be straightforward to add facial recognition to your flows. Having the module will unlock a setting inside the participant settings menu.
Here you can turn it on and off on a per flow basis. With this setting on, participants will then be required to identify themselves with a picture taken by the computer they are doing the assignment with. It will require this ID every time they start up they start the assignment on the flow.
Please note that facial recognition is not available for FLOWlock/FLOWmulti on iPad.
What does it look like for the participant?
When the participant enters a flow with facial recognition on, it will then be required to take a picture with either the integrated webcam on a laptop or by having an external webcam connected to the computer or laptop.
- Take Image: This is the button to press to take the image that will be used for the facial recognition of the participants. You can adjust the size of the window on your screen by using the dropdown menu to the left of it.
- Choose Webcam: This dropdown menu becomes available should the participant have multiple webcams available to take the image with. Note: External webcams have to plugged in before starting the flow, otherwise it will not register.
- Corner menu: By pressing this button, the participant gets 2 different options. They can either exit the browser on their own, or they can skip the facial recognition proces by having an invigilator use their password to allow the participant to skip.
After taking the photo, the participant will get the option to either retake the image or confirm this is the image they want to use. Doing so will add the face to an archive for facial recognition. The participants very first image will be used to cross-examine all future images taken for future flows.
Note: In case of multiple faces found on an image taken, facial recognition will make use of the largest face in the frame, which will almost always be the closest face as well, indicating which person to use for the facial recognition.
All images ever taken by the participant for facial recognition will be available to managers and invigilators on the same flow. This way the invigilators can access the images in case of a poor match or other, to make sure the participant is the real participant.
Furthermore, the FLOWmonitor will show all pictures taken during the exam in a timeline. All data and pictures are saved in the FLOWmonitor and it is possible to go through all the participants on the flow after the participation time has ended if you, for example, wish to follow up on those participants with a low general match score.
The first time a student enters with facial recognition on, they will be displayed as "New" in the flowmonitor. When a participant has skipped or left the exam before taking an image, they will get the "Not answered" on the monitor. The identification process can be skipped with the use of the invigilator password.
When it is not the first time a participant enters with facial recognition on, they will get a percentage based on how well the face on their latest image matches the first image ever taken. If a poor facial recognition happens, it is possible to open up the pictures that were used in the match, by clicking the button with the percentage.
In this window, the person monitoring can see why there might have been a bad match between the two pictures. Since only the face is used, different clothing or hairstyles will not affect the percentage given.
If the person monitoring the flow, is uncertain whether there might be signs of impersonating or alike, they can also see previous matches made, to see if major changes have happened since last time they used facial recognition. This dropdown will give you access to see every match since the very first reference ever made.
Disable facial recognition on participant level
(Released with minor release April 2020)
When the facial recognition module is enabled on the flow all participants need to identify themselves by using the webcam before starting the flow and their webcam takes pictures during the whole period of participation. But it is also possible to disable the use of facial recognition for one or multiple participants on the flow.
This is a task for managers and the setting is placed under All participation setting where the facial recognition module is enabled on the flow.
Note: Mr. Wiseflow and his little helper shown in this guide were used to replace the real people used for the facial recognition in the pictures. Mr. Wiseflow does not get recognised by the facial recognition software as having a face.