The Approach for Action Recognition Based on the Reconstructed Phase Spaces

Figure 4

Sample frames from our datasets. The action labels in each dataset are as follows: (a) KTH dataset [33]: walking (a1), jogging (a2), running (a3), boxing (a4), and hand clapping (a5); (b) Weizmann dataset [21]: bending (a1), jumping jack (a2), jumping forward on two legs (a3), jumping in place on two legs (a4), running (a5), galloping sideways (a6), walking (a7), waving one hand (a8), and waving two hands (a9); (c)UCF sports action dataset [37]: diving(a1), golf swinging (a2), kicking (a3), lifting (a4), horseback riding (a5), running (a6), skating (a7), swinging (a8), and walking (a9); (d) our action dataset: walking (a1), jogging (a2), running (a3), boxing (a4), and handclapping (a5).

(a)

(b)

(c)

(d)