Table 4.

$F_1$ score performance on OPPORTUNITY dataset for the gestures and modes of locomotion recognition tasks, either including or ignoring the Null class. The best results are highlighted in bold.

Method	Modes of Locomotion	Modes of Locomotion	Gesture Recognition	Gesture Recognition
	(No Null Class)		(No Null Class)
	OPPORTUNITY Challenge Submissions
LDA	0.64	0.59	0.25	0.69
QDA	0.77	0.68	0.24	0.53
NCC	0.60	0.54	0.19	0.51
1 NN	0.85	0.84	0.55	0.87
3 NN	0.85	0.85	0.56	0.85
UP	0.84	0.60	0.22	0.64
NStar	0.86	0.61	0.65	0.84
SStar	0.86	0.64	0.70	0.86
CStar	0.87	0.63	0.77	0.88
NU	0.75	0.53
MU	0.87	0.62
	Deep architectures
CNN [17]				0.851
Baseline CNN	0.912	0.878	0.783	0.883
DeepConvLSTM	0.930	0.895	0.866	0.915