Action Recognition by Joint Spatial-Temporal Motion Feature

Table 1

Experiment on Weizmann dataset.

Method		Recognition rate	Recognition time
Enhanced DTW	Original image size	100%	5.33 s
CFC-DTW	Level 1: 30% size; Level 2: 100% size; 𝐾=3	97.8%	1.17 s
	Level 1: 30% size; Level 2: 100% size; 𝐾=5	98.9%	1.30 s
	Level 1: 20% size; Level 2: 50% size; 𝐾=10	98.9%	0.55 s