Table 5 RF classification confusion matrix of 13 deciduous broadleaf species.

From: Assessing the potential of multi-seasonal Sentinel-2 satellite imagery combined with airborne LiDAR for urban tree species identification

 

SAMU6

SABA

ZESE80

GIBI2

KOPA

MADE7

PRSE3

PLAC

TRSE6

LIFO

PRCE2

ACRU

POAL7

Total

UA (%)

SAMU6

46

6

2

1

2

0

7

0

0

2

1

2

0

69

66.67

SABA

4

42

2

1

3

1

1

1

0

0

0

0

0

53

75.47

ZESE80

4

2

36

3

3

1

2

1

1

2

0

0

1

56

64.29

GIBI2

6

6

1

43

2

2

2

7

0

1

0

0

1

71

60.56

KOPA

3

7

4

1

38

0

3

3

0

1

1

0

1

62

61.29

MADE7

1

4

3

1

3

38

1

0

1

0

1

0

0

53

71.7

PRSE3

2

0

5

0

2

2

31

2

0

0

0

0

0

44

70.45

PLAC

2

0

2

1

3

1

1

35

0

1

0

1

1

48

72.92

TRSE6

4

5

4

2

3

2

0

1

12

0

1

0

0

34

35.29

LIFO

0

1

2

1

2

1

0

2

1

18

0

2

0

30

60

PRCE2

4

1

1

0

1

0

2

0

1

0

15

0

0

25

60

ACRU

1

1

2

2

1

0

4

2

0

2

1

11

0

27

40.74

POAL7

1

0

1

1

0

0

1

3

0

0

1

0

17

25

68

Total

78

75

65

57

61

48

55

57

16

27

21

16

21

OA (%) = 63.32

F1-score = 0.63

PA (%)

58.97

56

55.38

75.44

59.02

79.17

56.36

61.4

75

66.67

71.43

68.75

80.95

  1. The number of correctly classified samples is in bold. 
  2. PA producer’s accuracy, UA user’s accuracy, OA overall accuracy.