Table 9.

Experimental comparison of full-body re-identification algorithms (cross-domain Re-ID). Symbols: M—Market-1501; D—DukeMTMC-reID; MSMT17 Merged—MSMT17, which combines training and test parts.

Algorithm	MSMT17 Merged → M		MSMT17 Merged → D
	${\mathit{Rank}}_{\mathbf{1}}\mathbf{(}\mathbf{\uparrow }\mathbf{)}\mathbf{,}\mathbf{\%}$	$\mathit{mAP}\mathbf{(}\mathbf{\uparrow }\mathbf{)}\mathbf{,}\mathbf{\%}$	${\mathit{Rank}}_{\mathbf{1}}\mathbf{(}\mathbf{\uparrow }\mathbf{)}\mathbf{,}\mathbf{\%}$	$\mathit{mAP}\mathbf{(}\mathbf{\uparrow }\mathbf{)}\mathbf{,}\mathbf{\%}$
Baseline [10]	65.65	37.71	66.16	47.74
OSNet-IBN [19]	66.47	37.22	67.43	45.59
OSNet-AIN [61]	70.13	43.28	71.06	52.72
QAConv [21]	72.64	43.11	—	—
Our	71.08	43.39	72.51	53.71
Our + Filter Module	73.69	46.81	77.55	60.99

The best results by metrics are highlighted in bold.