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

Table 4

Performance comparison between backpropagation CNN model and feedforward proposed technique.

Deep spatiotemporal features	Validation Scheme	BP-CNN Accuracy %	Proposed technique; Accuracy %
x17	Splits	78.13	81.43
	LOAO	66.19	67.975
	LOSO	57.57	82.19
x 39	Splits	62.50	70.20
	LOAO	53.37	55.74
	LOSO	68.46	77.76