Table 10 Shows the mAcc values for multiple categories average among different low-light enhancement methods.

From: CUI-Net: a correcting uneven illumination net for low-light image enhancement

Method

RO

SI

BU

WA

FE

PO

TL

TS

VE

TE

SK

PE

RI

CA

TR

MO

BI

mAcc

Base

92.1

69.8

71.9

36.8

29.3

85.3

73.3

60.0

36.1

22.6

45.7

43.9

14.4

74.9

87.2

6.0

54.0

50.2

EnGAN

96.2

82.4

86.8

47.2

41.9

69.6

76.4

51.7

84.1

11.1

91.8

29.4

14.5

84.5

93.4

2.9

66.3

57.2

KinD

95.7

70.5

86.8

45.7

31.9

73.4

71.7

54.4

79.1

8.7

92.9

31.4

15.2

83.7

79.5

3.8

32.3

53.2

ZeroDCE

94.6

85.5

87.6

43.7

46.1

76.8

80.2

53.0

86.1

12.2

88.0

40.6

11.6

80.5

96.0

3.1

48.1

57.4

DCE++

95.5

69.8

85.6

32.7

55.5

72.3

72.9

46.8

77.1

34.9

85.0

24.2

3.5

70.1

87.2

5.2

49.7

53.8

RUAS

91.7

85.3

90.9

48.0

38.9

73.5

57.8

38.4

73.0

11.4

93.5

19.0

0.9

71.7

79.4

1.8

28.3

50.2

SCI

96.4

78.0

88.2

59.4

44.2

75.4

72.1

58.2

80.8

10.1

86.4

37.9

15.4

81.8

96.5

2.9

18.1

55.7

Ureti

95.1

85.4

87.4

42.6

37.3

61.8

68.5

56.4

81.3

9.7

94.7

34.5

9.8

78.1

94.6

2.7

34.3

54.1

CUI

96.9

84.3

87.8

56.5

48.5

74.3

76.1

65.4

82.2

23.8

90.8

48.2

38.3

79.2

95.5

18.7

55.9

62.4

  1. The symbol set {RO, SI, BU, WA, FE, PO, TL, TS, VE, TE, SK, PE, RI, CA, TR, MO, BI} represents {road, sidewalk, building, wall, fence, pole, traffic light, traffic sign, vegetation, terrain, sky, person, rider, car, train, motorcycle, bicycle}. CUI-Net had the highest mAcc value for four segmentation targets.