Multiple Feature Fusion Based on Co-Training Approach and Time Regularization for Place Classification in Wearable Video

Table 2

IMMED dataset: average accuracy of the multiple feature approaches.

Feature/approach	MKL	MKL-TA	DAS	DAS-TA	CO-DAS	CO-TA-DAS
BOVW-SPH	0.48	0.50	0.51	0.56	0.50	0.53
BOVW-CRFH	0.50	0.54	0.51	0.56	0.54	0.58
SPH-CRFH	0.48	0.51	0.50	0.54	0.54	0.57
BOVW-SPH-CRFH	0.48	0.51	0.51	0.56	—	—