Skip to main content

Table 2 Classification matrices of the LDA models for discrimination between non-aged (control) and aged seeds using spectral signatures (at 20 wavelengths) extracted from multispectral images of cowpea seeds

From: Utilization of computer vision and multispectral imaging techniques for classification of cowpea (Vigna unguiculata) seeds

Data set

Two-class classification

 

Five-class classification

  

Non-aged

Aged

% Correct

 

Non-aged

AA24

AA48

AA72

AA96

% Correct

Training (n = 401)

Non-aged

72

9

88.89%

Non-aged

75

3

3

0

1

91.46%

 

Aged

1

319

99.69%

AA24

1

77

3

0

0

95.06%

 

Overall correct classification

  

97.51%

AA48

0

0

57

3

12

79.17

     

AA72

0

1

7

71

6

83.53%

     

AA96

0

0

9

6

66

81.48%

      

Overall correct classification

    

86.28%

Cross-validation (n = 401)

Non-aged

69

12

85.19

Non-aged

71

6

3

0

2

86.59%

 

Aged

1

319

99.69

AA24

1

75

4

1

0

92.59%

 

Overall correct classification

  

96.76

AA48

0

2

53

3

14

73.61

     

AA72

0

1

9

67

8

78.82%

     

AA96

0

1

10

7

63

77.78%

      

Overall correct classification

    

82.04

Validation (n = 100)

Non-aged

16

3

84.21

Non-aged

17

1

0

0

0

94.44%

 

Aged

0

81

100

AA24

0

17

0

2

0

89.47%

 

Overall correct classification

  

97.0

AA48

0

1

20

2

5

71.43

     

AA72

0

0

1

15

0

93.75%

     

AA96

0

0

1

0

18

94.74%

      

Overall correct classification

    

87.00