Surface Defect Target Identification on Copper Strip Based on Adaptive Genetic Algorithm and Feature Saliency

Table 1

Identification accuracy rates using single feature.

Feature	Training set		Testing set
	Total correct rate	Target correct rate	Total correct rate	Target correct rate
f1	0.8056	0.7533	0.8554	0.7727
f2	0.5637	0.4801	0.5373	0.3359
f3	0.6119	0.5003	0.6883	0.4988
f4	0.6715	0.7092	0.7004	0.5013
f5	0.5508	0.3342	0.6017	0.3081
f6	0.7702	0.6547	0.8009	0.6833
f7	0.8113	0.7087	0.8351	0.7104
f8	0.7518	0.6476	0.7991	0.6506
f9	0.8683	0.7812	0.8790	0.7941
f10	0.6013	0.4747	0.5706	0.3987
f11	0.8896	0.8900	0.8711	0.9014
f12	0.9004	0.9151	0.9107	0.8992
f13	0.9417	0.9450	0.9625	0.9633
f14	0.8006	0.7229	0.8513	0.7094
f15	0.5725	0.3918	0.5631	0.3614