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

Table 3

Experimental results of Caltech-256 data set.

	AlexNet	VggNet	GoogLeNet	AA-ResNet	DenseNet
Train Acc	96.28%	98.47%	96.34%	99.88%	93.94%
Test Acc	40.13%	33.21%	35.67%	52.29%	47.26%
Loss	0.12786	0.04983	0.14234	0.00217	0.28646
Time (s)	152.34	176.92	209.25	211.46	263.55