IEEE Consumer Electronics Magazine - May/June 2022 - 28

Healthcare Technology
Figure 3. Monitoring IoT node prototype.
Figure 2. (a) Monitoring IoT node algorithm. (b)
Training code that runs on Colab.
The Internet connection of the monitoring
node can be achieved by a local Wi-Fi network, if
there is none, an extra GSM module can be used.
The node has an internal battery to operate in
case of a power outage. Eachmonitoring IoT node
is collected by authorized personnel when the
quarantine period of the relevant individuals
expires and placed in another quarantine home
after the disinfection process.When the pandemic
is over, these devices can be used to monitor the
attendance of students to courses at schools or
universities withminor changes in the software.
Experimental Study
The designed system prototype was used to
monitor the home quarantine of a family during
their real COVID-19 infection period. All members
of the family with two children, one 1-yearold
and the other 6-year-old living in the same
house were infected. The training was carried
out using a pretrained model with a total of 40
photographs, 10 photographs of each quarantine
subject. The magnetic switch IoT node is
placed at the exit door of the house and the monitoring
IoT node communicated over the local
28
Wi-Fi network they created. It was ensured that
all quarantine subjects had their body temperature
and faces scanned three times a day. Moreover,
due to the opening of the door of the home
7 times during the 10-day quarantine period, all
quarantine subjects were requested to perform
the scanning process within the 3 min after it
was opened. A total of 37 facial and body temperature
scans were performed during the 10-day
quarantine period.
According to the log file, the face recognition
process has 94% accuracy on the first try, only two
recognition processes resulted in the error. However,
the quarantine subject managed to introduce
herself to the system by not leaving the camera
after the first try. The body temperature of one of
the children was high once during the measurement
and thiswas written in the real-time database.
The prototype of the monitoring IoT node is
shown in Figure 3. Since the face recognition process
depends strongly on the ambient lighting,
an LED lighting system has been added to
increase the accuracy of the system, which is
not shown here. Because of the vulnerableness
of face recognition systems to the spoof attacks,
the captured images are transferred to storage.
For our future studies, the Liveness detection
algorithm that runs on another server can check
for the most recently uploaded images. If the picture
is real, it can be deleted from the storage;
otherwise, it can be reported as a violation.
CONCLUSIONS
Home quarantine of infected or close contact
individuals who are in relatively good health is one
IEEE Consumer Electronics Magazine
IEEE Consumer Electronics Magazine - May/June 2022

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - May/June 2022
