Table 8 Comparative experimental results of models for each defect type on the GC10-DET dataset.

From: DEENet: an edge-enhanced CNN–Transformer dual-encoder model for steel surface defect detection

Methods

mAP/%

Precision/%

Recall/%

Param/M

FLOPs/G

YOLOv5s

64. 3

68.1

53.2

7. 2

27.7

YOLOv9

68. 7

64.3

55.7

12. 1

32.9

YOLOv8

69.2

65.7

63.5

12.4

8.2

YOLOv10

70. 4

69.7

70.1

8. 0

40.6

YOLOv11

69. 7

69.2

68.2

9. 4

42.8

RT-DETR33

69.4

68.8

68.6

42

136

MSD-YOLO34

65.6

63.9

64.3

35.3

54.2

MD-YOLO35

69.3

69.0

70.4

9.0

14.1

DEENet

71.5

70.3

71.6

8.2

12.4

  1. Significant values are in bold.
Back to article page