Table 1 Different ablation experimental results of GEW-YOLO model on Dockship dataset. y/n represents none.

From: Multi-ship detection and classification with feature enhancement and lightweight fusion

Model

GSConvns

ESSE

Wise-IoU

Precision(%)

Recall(%)

mAP@0.5(%)

YOLOv8

n

n

n

80

70.9

78.3

YOLOv8+GSConvns

y

n

n

86

60

76.6

YOLOv8+ESSE

n

y

n

79.4

67.4

77.8

YOLOv8+Wise-IoU

n

n

y

80.6

62

71.1

y/n GSConvns

n

y

y

78.5

69.6

79.2

y/n ESSE

y

n

y

75.4

70.6

74.1

y/n Wise-IoU

y

y

n

79.4

67.4

75.6

GEW-YOLO(ours)

y

y

y

79.4

76.9

82.1

  1. Significant values are in [bold].