Skip to main content

Table 2 Race confusion tables

From: A study on different experimental configurations for age, race, and gender estimation problems

(a) LBP—unbalanced

 

(b) LBP—balanced

 
 

W T (%)

B T (%)

A T (%)

A T (%)

 

W T (%)

B T (%)

A T (%)

A T (%)

\({W_{P}^{N}}\)

95.7

2.1

0.5

1.7

\({W_{P}^{N}}\)

62.8

5

10

22.2

\({B_{P}^{N}}\)

0.8

98.8

0.1

0.3

\({B_{P}^{N}}\)

4.8

81.8

3.6

9.8

\({A_{P}^{N}}\)

22.6

10.3

57.3

9.8

\({A_{P}^{N}}\)

10.2

5

70.8

14

\({H_{P}^{N}}\)

31.9

14.2

2.7

51.2

\({H_{P}^{N}}\)

22.4

10.4

14.4

52.8

\({W_{P}^{S}}\)

96

2.2

0.3

1.5

\({W_{P}^{S}}\)

64.6

4.8

9.4

21.1

\({B_{P}^{S}}\)

0.8

98.8

0.1

0.2

\({B_{P}^{S}}\)

5.8

79.8

3.4

1.1

\({A_{P}^{S}}\)

22.1

11

55.9

11

\({A_{P}^{S}}\)

10

4.40

71.6

14

\({H_{P}^{S}}\)

32.5

15.9

2.2

49.4

\({H_{P}^{S}}\)

21.2

9.4

12.6

56.8

TA N

75.7

 

TA N

67.1

 

TA S

75

 

TA S

68.2

 

(c) HOG—unbalanced

 

(d) HOG—balanced

 
 

W T (%)

B T (%)

A T (%)

A T (%)

 

W T (%)

B T (%)

A T (%)

A T (%)

\({W_{P}^{N}}\)

93.9

6.1

0

0

\({W_{P}^{N}}\)

44.6

10

16.6

28.8

\({B_{P}^{N}}\)

0.8

99.2

0

0

\({B_{P}^{N}}\)

12.6

58.8

13.2

15.4

\({A_{P}^{N}}\)

42.9

57.1

0

0

\({A_{P}^{N}}\)

14.6

8.2

57

20.2

\({H_{P}^{N}}\)

57.7

42.3

0

0

\({H_{P}^{N}}\)

14.6

13

15.2

46

\({W_{P}^{S}}\)

94.3

3.6

0.3

1.8

\({W_{P}^{S}}\)

43.2

8.6

15.4

32.8

\({B_{P}^{S}}\)

1

98.6

0

0.3

\({B_{P}^{S}}\)

14.4

54

12.8

18.8

\({A_{P}^{S}}\)

20

12.1

57.4

10.5

\({A_{P}^{S}}\)

15.8

6.6

55.4

22.2

\({H_{P}^{S}}\)

30.5

17.3

2

50.2

\({H_{P}^{S}}\)

26.2

11.2

17

45.6

TA N

48.3

 

TA N

51.6

 

TA S

75.1

 

TA S

49.5

 

(e) SWLD—unbalanced

 

(f) SWLD–balanced

 
 

W T (%)

B T (%)

A T (%)

A T (%)

 

W T (%)

B T (%)

A T (%)

A T (%)

\({W_{P}^{N}}\)

94.6

3.4

0.4

1.6

\({W_{P}^{N}}\)

51.2

9.8

14

25

\({B_{P}^{N}}\)

1

98.5

0.1

0.4

\({B_{P}^{N}}\)

12.6

58.8

9.6

19

\({A_{P}^{N}}\)

23.5

10.8

56.9

8.8

\({A_{P}^{N}}\)

15

6.2

61.8

17

\({H_{P}^{N}}\)

37.8

19.9

2.4

39.9

\({H_{P}^{N}}\)

22.4

17.6

16.4

43.6

\({W_{P}^{S}}\)

94.3

3.4

0.4

1.9

\({W_{P}^{S}}\)

50.4

8.2

13.2

28.2

\({B_{P}^{S}}\)

1.1

98.5

0.1

0.3

\({B_{P}^{S}}\)

13.8

54.2

9.4

22.6

\({A_{P}^{S}}\)

25

10.6

53.2

11.2

\({A_{P}^{S}}\)

14.6

6.4

60.4

18.6

\({H_{P}^{S}}\)

36.6

20.2

2

41.2

\({H_{P}^{S}}\)

24.4

15.6

15.2

44.8

TA N

72.5

 

TA N

53.8

 

TA S

71.8

 

TA S

52.5

 

(g) CLBP—unbalanced

 

(h) CLBP—balanced

 
 

W T (%)

B T (%)

A T (%)

A T (%)

 

W T (%)

B T (%)

A T (%)

A T (%)

\({W_{P}^{N}}\)

96.3

1.7

0.4

1.6

\({W_{P}^{N}}\)

64.2

3.6

10.6

0

\({B_{P}^{N}}\)

0.6

99.1

0.1

0.2

\({B_{P}^{N}}\)

7.2

75

4.8

13

\({A_{P}^{N}}\)

15.6

7.6

65.2

11.6

\({A_{P}^{N}}\)

9.8

4.4

71.4

14.6

\({H_{P}^{N}}\)

27.4

12.4

2.9

57.3

\({H_{P}^{N}}\)

22.2

7.8

15.6

54.4

\({W_{P}^{S}}\)

96.3

1.8

0.3

1.6

\({W_{P}^{S}}\)

62.8

3.60

11.2

22.4

\({B_{P}^{S}}\)

0.6

99.1

0.2

0.2

\({B_{P}^{S}}\)

7.2

73.6

5

14.2

\({A_{P}^{S}}\)

15.3

8.6

61.6

14.5

\({A_{P}^{S}}\)

10.4

4

70

15.6

\({H_{P}^{S}}\)

27.3

13.1

2.4

57.2

\({H_{P}^{S}}\)

22.6

8.4

14.6

54.4

TA N

79.4

 

TA N

66.3

 

TA S

78.5

 

TA S

65.2

 
  1. Each table presents the results for a specific descriptor for both scaled and non-scaled approaches. Moreover, the tables in the left column have been trained on the unbalanced dataset; the tables in the right column over the balanced one. White, Black, Asian, and Hispanic classes are referred respectively with W, B, A, and H. Subfix P and T indicate predicted and true value, respectively. TA is for total accuracy; superfix N and S refer to the non-scaled and scaled approaches, respectively