An Easy Way to Deploy the Re-ID System on Intelligent City: Cross-Domain Few-Shot Person Reidentification

Table 2

Comparison results for different methods under a single source set. All settings are experimented repeatedly for 10 times and get the average value.

Source —> Target	Size	Policy1		Policy2		Policy3		ReWIF_S		ReWIF
		mAP	Rank-1	mAP	Rank-1	mAP	Rank-1	mAP	Rank-1	mAP	Rank-1
Duke—>Market	50	21.2	47.9	78.3	89.5	79.7	91.7	81.9	93.1	82.2	95.9
Duke—>Market	10	18.5	43.2	61.1	77.7	76.2	90.3	80.4	91.7	81.9	94.8
Duke—>Market	5	15.5	44.4	46.9	62.5	65.5	82.3	68.6	85.6	69.9	88.6
Duke—>Market	1	13.4	40.0	32.4	58.6	61.2	80.4	66.5	81.2	67.2	85.9
Market—> Duke	50	19.3	40.8	66.0	79.9	71.7	89.7	76.7	91.9	79.6	93.4
Market—> Duke	10	18.2	40.0	43.1	68.6	70.2	87.1	75.8	87.1	77.9	91.1
Market—> Duke	5	16.8	37.1	31.9	58.8	63.2	80.1	65.9	81.9	74.8	87.2
Market—> Duke	1	15.4	33.3	27.4	46.6	60.5	75.9	62.8	79.3	70.1	84.1