IIPA-Net: Joint Illumination-Invariant and Pose-Aligned Feature Learning for Person Reidentification

Table 3

Ablation study on normal- and low-light market datasets.

Condition	Method	Rank-1	mAP
Normal	Ours w/o attention	94.4	89.5
	Ours w/o retinex	92.6	88.3
	Ours	96.2	90.3
Low light	Ours w/o attention	44.7	17.3
	Ours w/o retinex	56.3	22.4
	Ours	60.5	27.7

w/o: without; bold: best results.