An Image Classification Method Based on Adaptive Attention Mechanism and Feature Extraction Network

Table 2

Experimental results of Caltech-101 data set.

	AlexNet	VggNet	GoogLeNet	AA-ResNet	DenseNet
Train Acc	98.45%	99.29%	99.35%	99.96%	99.81%
Test Acc	70.07%	63.74%	63.32%	69.61%	71.28%
Loss	0.05628	0.03123	0.01042	0.00352	0.03112
Time (s)	42.55	43.56	61.27	55.82	70.91