Table 2.

Experimental results on UTD-MHAD.

Technic Used	Accuracy of Recognition	Characteristics
Joint Trajectory Map	89.81%	Accumulating the user’s joints trajectory, Based on CNN [17].
Joint Distance Maps	88.1%	The distance between the joints of the user’s expression, based on CNN [16].
Motion History Map	84.00%	Cumulative recording of the user’s movement by the use of an image, CNN-based [15].
Depth Motion Map	79.10%	Human action recognition using a depth camera and a wearable inertial Sensor. Data fusion from multi-sensors [27].
Proposed Approach	99.40%	Differential evolution to optimize weights. DTW Based.