Motion Recognition Based on Deep Learning and Human Joint Points

Table 5

Comparison of quantitative effects of various models.

Model	Precision: raw	Precision quantization	Delay: initial	Delay quantization (ms)	Size: initial (MB)	Size quantization (MB)
MobileNetV1	0.708	0.654	124 ms	112	16.6	4.3
MobileNetV2	0.709	0.632	89 ms^3	98	13	3.6
InceptionV3	0.77	0.762	1130 ms	845	95.9	23.9
ResnetV2	0.76	0.758	973 ms	2868	178.5	44.9