Skip to main content

Table 1 Performance of the R-net model for male detection in the testing set

From: Detecting representative characteristics of different genders using intraoral photographs: a deep learning model with interpretation of gradient-weighted class activation mapping

Group

Actual Male

Actual Female

Sensitivity

Specificity

True Positive

False Negative

True Negative

False Positive

Uncropped Children Group

44

2

28

19

0.957

0.698

Cropped Children Group

42

4

29

18

0.913

0.7

Uncropped Youth Group

52

1

49

5

0.981

0.912

Cropped Youth Group

50

3

44

10

0.943

0.833

Uncropped Testing Samples

96

3

77

24

0.970

0.762

Cropped Testing Samples

92

7

73

28

0.929

0.723