IEEE Consumer Electronics Magazine - March/April 2021 - 96

Consumer Technology-Based Solutions for COVID-19

Table 2. Classification of Number of Persons Present for
Static and Dynamic Crowd Scenario.

corresponding accuracy achieved for the static
and dynamic crowd cases are 86:02% and
84:55%, respectively.

Testing
images

Accuracy
(%) for static
crowd

Accuracy (%)
for dynamic
crowd

200

250

85.9

84.35

CONCLUSION

300

350

86.3

85.36

400

500

86.02

84.55

500

700

85.3

83.02

600

800

85.6

83.72

750

1000

85.6

83.91

This article proposes a passive nonintrusive
solution for outdoor crowd detection and subsequently, its size estimation. The feasibility of
this novel approach was verified with practical
signal captured using SDR-based prototype
developed by the authors. The results prove the
feasibility of our proposal.
Consistency of the performance of this proposal was evaluated by calculating the detection
accuracy for static and dynamic crowd on a different day with a different set of people. Similar
performance was observed, which proves the
performance consistency.
For the static as well as dynamic crowd
cases, the nearest neighbor classifier provided
acceptable performance.
The analysis in this article with practical data
confirms that LTE-CommSense principle can successfully detect crowd in outdoor environment.
After detection, it can estimate the crowd size as
well with reasonable accuracy.

Training
images

Table 3. Confusion Matrix for Static and Dynamic Crowd
for the Case of 500 Training Data per Category and 700
Testing Data per Category.

& REFERENCES
1. V. S. Rathore, V. Kumawat, and B. Umamaheswari,
" Coronavirus (COVID-19) in India-Statistics &facts, "
in Proc. 4th World Conf. Smart Trends Syst., Secur.,
Sustainability, 2020, pp. 518-521.

size, i.e., the number of persons in the crowd.
Table 2 depicts the evaluated accuracy for different numbers data for training and size estimation. A simple nearest neighbor classifier was
chosen for this purpose. When new sample data
are input to the system, using the nearest neighbor classifier the distance of that from the nearest data available in the training dataset
was evaluated. The results reveal that the proposed approach is a promising candidate for
crowd detection and its size estimation in an
outdoor environment consisting of either static
or dynamic persons present in the crowd.
Table 3 shows the confusion matrix for the
lowest accuracy achieved in Table 2. The

2. T. Coughlin, " Impact of COVID-19 on the consumer
electronics market, " IEEE Consum. Electron. Mag.,
to be published, doi: 10.1109/MCE.2020.3016753.
3. R. Abbas and K. Michael, " COVID-19 contact trace
app deployments: Learnings from Australia and
Singapore, " IEEE Consum. Electron. Mag., vol. 10,
no. 5, pp. 65-70, Sep. 2020.
4. K. Michael and R. Abbas, " Behind COVID-19 contact
trace apps: The Google-Apple partnership, " IEEE
Consum. Electron. Mag., vol. 9, no. 5, pp. 71-76,
Sep. 2020.
5. " Multilateral call for solutions for COVID-19 echo
period - life without a vaccine. " 2020. Accessed: Jun. 19,
2020. [Online]. Available: https://www.eurekanetwork.org/
content/multilateral-call-solutions-covid-19-echo-period

IEEE Consumer Electronics Magazine

https://www.eurekanetwork.org/content/multilateral-call-solutions-covid-19-echo-period https://www.eurekanetwork.org/content/multilateral-call-solutions-covid-19-echo-period

IEEE Consumer Electronics Magazine - March/April 2021

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - March/April 2021