Table 3.

Grasping evaluation of detection using a modified YOLOv2 and another two.

Class of Product	Method	Parameters
		Confid.	Accur.	Precis.	Recall	F1	AP	Time (s)
A	modYOLOv2	0.942	0.990	0.990	1	0.995	0.990	0.055
	MobileNet2	0.905	0.963	0.963	1	0.981	0.941	0.157
	ResNet18	0.890	0.950	1	1	1	0.961	0.056
B	modYOLOv2	0.938	0.972	0.971	1	0.985	0.927	0.054
	MobileNet2	0.626	0.925	1	1	1	0.958	0.053
	ResNet18	0.707	0.750	0.944	1	0.971	0.971	0.056
C	modYOLOv2	0.852	0.990	1	0.989	0.994	0.771	0.053
	MobileNet2	0.768	0.740	1	0.958	0.978	0.827	0.087
	ResNet18	0.719	0.750	0.969	0.969	0.969	0.633	0.053
D	modYOLOv2	0.925	0.990	0.990	1	0.995	0.990	0.054
	MobileNet2	0.864	0.963	0.963	1	0.981	0.909	0.053
	ResNet18	0.862	0.925	0.974	1	0.987	0.923	0.053
E	modYOLOv2	0.885	0.854	0.854	1	0.921	0.754	0.053
	MobileNet2	0.818	0.777	0.777	1	0.875	0.650	0.054
	ResNet18	0.709	0.875	0.875	1	0.933	0.739	0.052
F	modYOLOv2	0.922	0.981	0.981	1	0.990	0.979	0.054
	MobileNet2	0.708	0.963	0.963	1	0.981	0.998	0.053
	ResNet18	0.690	0.875	0.875	1	0.933	0.739	0.052
G	modYOLOv2	0.875	0.945	0.990	0.954	0.971	0.867	0.055
	MobileNet2	0.641	0.851	1	0.851	0.920	0.776	0.058
	ResNet18	0.785	0.975	0.975	1	0.987	0.957	0.056
H	modYOLOv2	0.952	0.981	1	0.981	0.990	0.907	0.056
	MobileNet2	0.841	0.851	1	0.923	0.960	0.749	0.062
	ResNet18	0.952	0.925	0.974	1	0.987	0.925	0.058
I	modYOLOv2	0.917	1	1	1	1	1	0.056
	MobileNet2	0.868	0.925	1	1	1	0.9616	0.053
	ResNet18	0.742	0.950	0.974	0.974	0.974	1	0.055
µ modYOLOv2		0.912	0.967	0.975	0.992	0.982	0.909	0.054
µ MobileNet2		0.782	0.884	0.963	0.970	0.964	0.863	0.070
µ ResNet18		0.784	0.886	0.951	0.994	0.971	0.872	0.055
σ modYOLOv2		0.032	0.043	0.044	0.015	0.015	0.089	0.001
σ MobileNet2		0.097	0.079	0.068	0.049	0.039	0.112	0.032
σ ResNet18		0.089	0.079	0.043	0.012	0.022	0.124	0.002