Table 4 Comparison with state-of-the-art models on the visdrone dataset.

From: Multi-scale feature fusion and feature calibration with edge information enhancement for remote sensing object detection

Model

mAP

AP50

AP75

APS

APM

APL

Para

GFLOPs

FPS

Baseline

26.0

43.8

26.2

18.1

36.4

43.2

19.88 M

56.9

61.4

Faster R-CNN

17.1

34.2

16.3

11.7

28.5

32.6

41.70 M

251.2

15.2

YOLOv5l

24.9

41.2

25.3

14.9

37.7

43.1

53.14 M

134.7

50.8

YOLOv6m

21.7

36.5

21.9

12.0

34.2

44.2

34.90 M

85.8

69.5

YOLOv7x

26.8

46.5

26.5

17.9

37.9

41.4

70.88 M

189.1

24.11

YOLOv8m

23.4

39.3

23.4

14.0

35.8

42.4

25.85 M

78.7

75.4

YOLOv9m

24.1

40.1

24.6

13.7

37.4

47.9

32.57 M

130.8

60.98

YOLOv10m

23.3

38.9

23.8

13.6

35.8

43.8

16.46 M

63.50

68.8

YOLOv11l

25.1

41.5

25.9

15.4

37.7

46.9

25.29 M

86.60

78.0

D-fine-m

25.7

39.6

23.4

15.8

32.7

42.3

19.19 M

56.37

46.9

Deformbale-DETR

25.4

44.0

24.8

17.1

35.5

38.6

40.96 M

173.0

34.0

RTDETR-fasternet

24.8

42.0

24.7

17

35

40.2

21.53 M

54.9

60.9

RTDETR- mobilenetv4

22.5

38.4

22.3

14.8

32

36.3

11.32 M

39.5

56.0

Our method

28.0

46.6

28.6

19.6

39

43.8

15.86 M

56.9

58.4

Back to article page