Table 3 Comparison of flops and parameters among different models.

From: Significant feature suppression and cross-feature fusion networks for fine-grained visual classification

Method

Baseline Model

Dataset

Accuracy rate (%)

Flops

Parameter

Cross-X

Resnet50

CUB-200-2011

87.7

32.16 GMac

48.31 M

MGE-CNN

Resnet50

CUB-200-2011

88.5

71.83 GMac

108.52 M

PMG

Resnet50

CUB-200-2011

89.3

37.42 GMac

45.13 M

FBSD

Resnet50

CUB-200-2011

89.0

47.33 GMac

44.46 M

Ours

Resnet50

CUB-200-2011

89.6

47.1G Mac

43.09 M

  1. Significant values are in bold.
Back to article page