Table 2 Performance comparison of YOLOv5s with different backbone networks.

From: YOLO-Granada: a lightweight attentioned Yolo for pomegranates fruit detection

Model(YOLO)

Params

GFLOPs

Size (MB)

F1 (%)

mAP@0.5

v5s(CSPDarknet-53)

7,033,114

15.9

13.76

88.9

0.928

MobileNetv2

4,603,738

10.0

9.20

87.1

0.917

MobileNetv3-large

5,583,072

10.8

11.05

88.5

0.92

MobileNetv3-small

3,586,312

6.5

7.20

87.3

0.911

EfficientNetv1

5,760,722

11.4

11.41

88.5

0.928

EfficientNetv2

27,298,066

35.8

52.74

89.6

0.933

ShuffleNetv2

3,803,738

8.1

7.65

87.1

0.916