Learning Identity-Consistent Feature for Cross-Modality Person Re-Identification via Pixel and Feature Alignment

Table 1

Comparison of CMC (%) and mAP (%) performances with other methods on SYSU-MM01.

Methods	All-search								Indoor-search
	Single-shot multishot								Single-shot multishot
	R1	R10	R20	mAP	R1	R10	R20	mAP	R1	R10	R20	mAP	R1	R10	R20	mAP
Two-stream [10]	11.65	47.99	65.50	12.85	16.33	58.35	74.46	8.03	15.60	61.18	81.20	21.49	22.49	72.22	88.61	13.92
One-stream [10]	12.04	49.68	66.74	13.67	16.26	58.14	75.05	8.59	16.94	63.55	82.10	22.95	22.62	71.74	87.82	15.04
Zero-padding [10]	14.80	54.12	71.33	15.95	19.13	61.40	78.41	10.89	20.58	68.38	85.79	26.92	24.43	75.86	91.32	18.86
BCTR [13]	16.20	54.90	71.5	19.2	—	—	—	—	—	—	—	—	—	—	—	—
BDTR [13]	17.1	55.5	72.0	19.7	—	—	—	—	—	—	—	—	—	—	—	—
D-HSME [26]	20.7	62.8	78.0	23.2	—	—	—	—	—	—	—	—	—	—	—	—
cmGAN [6]	27.0	67.5	80.6	27.8	31.5	72.7	85.0	22.3	31.7	77.2	89.2	42.2	37.0	80.9	92.3	32.8
DGD + MSR [12]	37.35	83.40	93.34	38.11	43.86	86.94	95.68	30.48	39.64	89.92	97.66	50.88	46.56	93.57	98.8	40.08
JSIA-ReID [20]	38.10	80.70	89.90	36.90	45.10	85.70	93.80	29.50	43.80	86.20	94.20	52.90	52.70	91.10	96.40	42.70
Ours	40.83	83.40	92.38	39.84	48.13	86.0	93.67	32.54	44.0	86.8	94.87	52.96	53.40	90.52	95.70	44.35