Learning a Mid-Level Representation for Multiview Action Recognition

Table 1

Action recognition accuracy comparison with state-of-the-art methods for both multiview action recognition and single-view action recognition on the IXMAS dataset.

Method	View 1	View 2	View 3	View 4	View 5	Average	Multiview
Junejo et al. [26]	76.4%	77.6%	73.6%	68.8%	66.1%	72.5%	72.7%
Weinland et al. [27]	85.8%	86.4%	88.0%	88.2%	74.7%	84.6%	83.5%
Wu et al. [28]	81.9%	80.1%	77.1%	77.6%	73.4%	78.0%	–
Chaaraoui et al. [18]	–	–	–	–	–	–	85.9%
Liu et al. [25]	84.5%	84.7%	88.0%	82.9%	83.4%	84.7%	88.0%
Pehlivan and Forsyth [11]	81.3%	86.3%	86.3%	85.9%	77.6%	83.5%	72.5%
Aryanfar et al. [29]	–	–	–	–	–	–	88.1%
Hsu et al. [17]	–	–	–	–	–	–	81.8%
Chun and Lee [15]	–	–	–	–	–	–	83.0%
Wang et al. [30]	–	–	–	–	–	–	84.9%
Sargano et al. [20]	–	–	–	–	–	–	89.7%
Our method	89.6%	88.4%	85.8%	78.2%	86.9%	85.8%	90.3%