Specificity of the model
The specificity of the model shows the rate in which an actual negative value is predicted correctly as a negative value. The specificity
scale is in percent. The specificity can be calculated by dividing the true negative value by the sum of true negative and false positives
value as seen in Equation 3. The results yielded 74.02% from using the data from Table 6. The result indicates that the model is able
to differentiate faces, but it is not accurate enough to be implemented as a biometric security system as it still has a significant false
positive prediction rate.
Conclusions
Several conclusions can be drawn from the research conducted. Firstly, the haar cascade method has a potential to make a facial
recognition system despite its high error rate with the use of the pre-trained classifier. Secondly, the accuracy of the model which is
74.8% shows that the model is good at recognizing faces but has trouble differentiating between faces from the system and outside
the system. Thirdly, the sensitivity of the model yields the value of 1, which means it can differentiate between the faces in the system
when shown just the faces in the system. Lastly, the specificity of the model yields 74.02% which indicates that the model can
differentiate between faces inside and outside of the system but not accurate enough to be implemented on a biometric securit y
system. Future works will include making a custom classifier and an integration of an expression recognition model as well.
151 |
P a g e
Conflicts of interest
The authors declare that there is no conflict of interest regarding the publication of this paper.
References
[1]
K. Schwab,
The Fourth Industrial Revolution; World Economic Forum: Geneva, Switzerland; ISBN 9781944835002.
, vol.
4, no. 3. 2016.
[2]
J. Woodward, C. Horn, J. Gatune, and A. Thomas,
Biometrics A Look at Facial Recognition
. 2003.
[3]
P. Viola and M. Jones, “Rapid object detection using a boosted cascade of simple features,”
Proc. IEEE Comput. Soc.
Conf. Comput. Vis. Pattern Recognit.
, vol. 1, no. July 2014, 2001, doi: 10.1109/cvpr.2001.990517.
[4]
M. A. Turk and A. P. Pentland, “Face recognition using eigenfaces,”
J. Cogn. Neurosci.
, vol. 3, 1991.
[5]
“Raspberry Pi 4,” 2021. https://www.raspberrypi.org/products/raspberry-pi-4-model-b/ (accessed Jul. 29, 2021).
[6]
C. Riggio, “What’s the deal with Accuracy, Precision, Recall and F1?,” 2019. https://towardsdatascience.com/whats-the-
deal-with-accuracy-precision-recall-and-f1-f5d8b4db1021 (accessed Aug. 15, 2021).
[7]
K. J. Bhanushree and M. B. Meenavathi, “Feature Based Face Recognition using Machine Learning Technniques,”
Int. J.
Recent Technol. Eng.
, vol. 8, no. 6, pp. 1313
–1317, 2020, doi: 10.35940/ijrte.f7497.038620.
[8]
M. Rovai, “Real-Time Face Recognition: An End-To-End Project,” 2018. https://www.raspberrypi.org/products/raspberry-
pi-4-model-b/ (accessed Mar. 30, 2021).
|
