Table5 Training and testing time and parameter comparison between different models.

From: Transformer based on channel-spatial attention for accurate classification of scenes in remote sensing image

Method

UCM (50%)

Accuracy

Train (s/epoch)

Test (s/epoch)

Parameters (M)

Flops (G)

ResNet-10137

92.47

16.1

6.9

46.0

7.6

ResNet-15210

92.95

23.5

9.3

60.0

11.0

SE-Net54

95.38

49.7

23.6

146.0

42.0

ViT-Base26

93.57

25.9

10.4

86.4

17.5

CSAT (our)

95.72

25.3

10.19

85.99

16.88

Back to article page