Table 7 Experimental results on the NEU-DET dataset.

From: Aluminum surface defect detection method based on a lightweight YOLOv4 network

Network model

[email protected]

cr

in

pa

ps

rs

sc

YOLOv5s

82.8

60.8

62.7

83.8

92.9

63.6

69.4

YOLOv4

82.3

66.8

56.6

87.4

96.5

57.3

54.5

YOLOv7

74.9

76.3

63.4

81.3

78.3

82.4

95.5

PPYOLO

79.7

55.1

51.9

77.3

87.6

49.5

58

Zhao’s method28

81.1

52.9

85.9

94.4

86.2

70.7

96.6

Yu’s method29

72.8

15.1

26.3

42.4

33.8

27.2

30.6

M2-BL-YOLOv4

84.3

75.5

72.8

87.2

97.0

79.9

93.4

Back to article page