Table 9 Generalization experiment on crack-seg dataset.

From: YOLOv8-seg-CP: a lightweight instance segmentation algorithm for chip pad based on improved YOLOv8-seg model

Method

mAP50 (Box)

mAP50 (Mask)

FLOPs

Params (M)

FPS

Model size (M)

inference time (ms)

YOLOv5n-seg

0.78

0.643

7.1G

2

217.47

4.2

4.6

YOLOv5s-seg

0.81

0.66

25.7G

7.3

146.711

15.2

6.81

YOLOv8n-seg

0.805

0.662

12.0G

3.4

876.9

6.8

1.14

YOLOv8s-seg

0.811

0.677

42.4G

11.7

531.8

23.9

1.88

YOLOv9c-seg

0.816

0.661

157.6G

27.6

209.7

56.2

4.77

Ours

0.823

0.706

8.2G

1.7

1143.9

3.7

0.87

Back to article page