Table 2 Model performance comparison.

From: CAFE-YOLO: an object detection algorithm from UAV perspective fusing channel attention and fine-grained feature enhancement

Model

Params (M)

GFLOPs

Inference time

P (%)

R (%)

mAP50 (%)

mAP50:95 (%)

Base

18.8

105.3

41.4

53.4

44.8

43.5

24.1

Base+CAG

19

108.5

43.7

53.5

45.1

43.8

24.2

Base+CAG+FGFE

8.23

103.7

37.8

52.7

46.7

44.2

24.1

CAFE-YOLO

8.62

102.6

31.9

53.7

46.3

44.6

24.5

Back to article page