Instructor Activity Recognition through Deep Spatiotemporal Features and Feedforward Extreme Learning Machines

Table 2

Performance of the proposed technique at the different network on the IAVID-I dataset using LOSO validation scheme.

Deep spatiotemporal features	Deep CNN layers	Features; Dimension	Hidden Nodes	LOSO Accuracy %
x 17	17	1x 4096	500	79.75
			2100	82.19
			5000	81.62
			10000	81.37
x 39	39	1x 4096	500	64.93
			2100	77.76
			5000	76.43
			10000	74.96