Table 4. Experimental results of VGG on ImageNet100.

Strategy	Pooling	Attn.	ImageNet100
			Acc.1 (%)	Acc.2 (%)
VGG_A	max pooling	N	71.16	71.67
VGG_B	FMAPooling	N	71.69	72.98
VGG_C	FMMPooling	N	70.82	72.09
VGG_D	FMAPooling	FMAttn	76.22	77.08
VGG_E	FMAPooling	CAM	74.01	74.98

Notes:

Acc.1: training with 155 epochs once.

Acc.2: training with 155 epochs twice.

The best results are highlighted in bold font.