Ensemble Learning-Based Person Re-identification with Multiple Feature Representations

Table 2

Matching rates of different approaches on the ETHZ dataset.

Rank
	1	5	10	20	1	5	10	20
Our approach	0.852	0.958	0.981	0.994	0.918	0.987	0.997	0.999
RS-KISS	0.770	0.921	0.962	0.985	0.835	0.963	0.984	0.996
RDC	0.690	0.858	0.922	0.969	0.727	0.901	0.956	0.988
LFDA	0.725	0.897	0.949	0.981	0.761	0.912	0.965	0.993
MFA	0.672	0.860	0.922	0.970	0.721	0.897	0.951	0.990
Adaboost	0.656	0.840	0.905	0.956	0.692	0.878	0.935	0.980
Bhat	0.555	0.761	0.840	0.906	0.610	0.809	0.878	0.941
PLS	0.483	0.694	0.780	0.868	0.546	0.751	0.833	0.924
Xing’s	0.544	0.752	0.833	0.904	0.608	0.803	0.874	0.936