Skip to main content
. 2015 Aug 5;9:103. doi: 10.3389/fncom.2015.00103

Figure 3.

Figure 3

Human motion samples for the six types of behavior in the KTH visual discrimination task. This illustration consists of different behaviors from a single person, while the whole data set contains 25 persons. Top row: Still frames from example video samples; boxing, clapping, waving, walking, running and jogging. Bottom row: Features extracted corresponding to the samples above, according to Equations (13) and (14). Features are the raw time-series activity used as input to the neural network.