150 |
P a g e
Table 6. Average of each category.
Values
1
st
iteration
2
nd
iteration
3
rd
iteration
4
th
iteration
5
th
iteration
Average of
each value
True Positive 3
3
3
3
3
3
False
Positive
25
20
28
30
23
25.2
False
Negative
0
0
0
0
0
0
True
Negative
72
77
69
67
74
71.8
Accuracy of the model
The datasets used for the training were the 100 grayscale images of the 3 unique faces. The accuracy of the model shows the rate
of correct predictions it has made over the total number of shown face. The scale of the accuracy is in percent. The accuracy can be
calculated by adding the true positive or TP and true negative or TN then dividing them with the total value which is 100 as seen in
Equation 1. The values are taken from Table 6 in which the obtained result will be 74.8%. This value indicates that the model has
decent accuracy for recognizing the shape of faces as seen that the model has 0 false negatives, but it has a difficulty in differentiating
between individual faces in the system as seen that there are an average of 25.2 false positives.
Sensitivity of the model
The sensitivity of the model shows the rate in which the actual positive value is predicted correctly as positive.
The scale of the
sensitivity is from 0 to 1.0, where 0 is the lowest and 1.0 is the highest. The sensitivity can be calculated by dividing the total true
positive value by the sum of true positive and false negative value as seen in Equation 2. The values from Table 6 will be used to
calculate the sensitivity in which the obtained value will be 1. This value indicates that the model can differentiate the 3 faces in the
database accurately, but not very accurate when shown with faces outside the database.
