The new data sets, i.e., pancake and sandwich making present different levels of complexity, since they involve several combinations of the basic activities under different objects. Furthermore, some tasks are performed in parallel using both hands while executing different activities. The human motions are captured by three cameras located in different positions, see Experimental Set-ups. These new recordings contain information of three external cameras, one mounted camera and one gaze camera.

Our data sets presents the following modalities to analyzed the visual information: