Automatic Human Gait Imitation and Recognition in 3D from Monocular Video with an Uncalibrated Camera

Table 4

Associated comparisons of the sole SVM or HCRF classifier on gait recognition at same window size 𝑤 and different viewing angles.

	Types
Viewing angle	Just after acquiring normalized parameters of real detected object						Just after reducing dimensionality of associated time sequences
	HCRF1 (𝑤=1)			SVM1			HCRF2 (𝑤=1)			SVM2
	Recognized rate (%)
	Thin coat	Thick coat	Backpack	Thin coat	thick coat	Backpack	Thin coat	Thick coat	Backpack	Thin coat	Thick coat	Backpack
54°	63.0	58.2	55.7	69.3	64.5	60.9	73.2	70.3	61.2	69.3	67.1	55.2
72°	64.2	60.5	54.8	72.1	65.5	55.4	75.4	70.5	62.4	72.2	65.3	58.1
90°	66.1	61.3	56.7	75.3	67.8	58.5	78.3	72.5	65.5	75.1	66.2	58.6
108°	73.4	67.5	55.5	74.9	69.0	58.9	80.8	78.1	66.8	78.0	66.3	59.9
126°	65.5	61.6	57.2	68.4	64.2	56.5	74.4	71.9	64.3	70.2	63.0	56.4