Table 2 Detection values on the GC10-DET dataset.

From: RSTD-YOLOv7: a steel surface defect detection based on improved YOLOv7

Methods

AP(%)

mAP(%)

P(%)

R(%)

FPS

pu

wl

cg

ws

os

ss

in

rp

cr

wf

YOLOv7

93.8

73.2

96.2

82.8

47.9

58.5

27.4

8.39

66.2

81.9

63.6

67.2

61.3

34.8

mothod 1

93.5

74.6

96.1

83.1

50.2

58.2

26.8

25.4

67.8

85.1

66.1

63.5

62.2

35.4

mothod 2

94.6

82.3

96.7

84.2

56.7

58.5

25.9

29.4

66.9

86.3

68.2

68.2

65.3

35.1

mothod 3

95.8

94.4

96.3

85.3

64.5

58.9

26.4

30.2

67.7

88.6

70.8

70.1

68.1

34.6

mothod 4

97.6

96.3

93.8

87.1

67.6

61.5

26.7

32.4

70.3

88.7

72.2

65.4

72.6

36.9

RSTD-YOLOv7

98.2

96.4

94.2

84.6

68.1

63.4

26.1

41.3

70.2

89.4

73.2

67.4

71.8

38.8

  1. Significant values are in bold.