An Integrated Model for On-Site Teaching Quality Evaluation Based on Deep Learning

Table 5

Performance comparison of each algorithm.

Model	Backbone network	FPS	AP (%)	Parameters	GMACS
YOLO3	Darknet-53	46	91.65	61.52M	27.91
YOLO4	CSPDarknet53	30	92.4	63.94M	25.46
YOLO5-S	Focus+CSP	47	91.9	7.06M	2.95
YOLO5-M	Focus+CSP	38	92.8	21.06M	9.07
YOLO5-L	Focus+CSP	29	93.8	47.06M	20.81
YOLO5-X	Focus+CSP	23	94.2	87.78M	39.43
YOLOX-S	Focus+CSP	45	93.1	8.94M	4.79
YOLOX-M	Focus+CSP	34	93.6	25.28M	13.23
YOLOX-L	Focus+CSP	27	94.1	54.21M	28.02
YOLOX-X	Focus+CSP	23	94.5	99.07M	50.75