A New Multiface Target Detection Algorithm for Students in Class Based on Bayesian Optimized YOLOv3 Model

Table 6

Test results of self-built classroom dataset.

Algorithm	45 students				100 students
	Accurate number	Accuracy rate (%)	Detection speed (Fps)	Parameter (MB)	Accurate number	Accuracy rate (%)	Detection speed (Fps)	Parameter (MB)
Fast R-CNN	32	86.49	4.92	57.3	66	88.00	9.85	93.5
FaceNet	33	89.19	7.35	55.9	70	93.33	12.21	95.1
DeepID	34	91.89	6.56	56.1	72	96.00	11.52	91.7
SSD	34	91.89	5.82	53.2	69	92.00	11.28	87.5
YOLOv1	32	86.49	6.23	56.6	71	94.67	11.32	92.3
YOLOv2	33	89.19	5.37	54.5	70	93.33	11.15	89.4
YOLOv3	34	94.59	5.12	55.3	72	96.00	10.59	90.6
Ours	36	97.30	5.35	54.2	73	97.33	11.26	90.1