Fine-Grained Cross-Modality Person Re-Identification Based on Mutual Prediction Learning

Shuang Li; Huafeng Li; Fan Li

doi:10.3788/LOP202259.1010010

Journals >Laser & Optoelectronics Progress >Volume 59 >Issue 10 >Page 1010010 > Article

Laser & Optoelectronics Progress
Vol. 59, Issue 10, 1010010 (2022)

Fine-Grained Cross-Modality Person Re-Identification Based on Mutual Prediction Learning

Shuang Li^1、2, Huafeng Li^1、2, and Fan Li^1、2、*

Author Affiliations

¹Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming 650500, Yunnan , China

²Yunnan Key Laboratory of Artificial Intelligence, Kunming 650500, Yunnan , China

show less

DOI: 10.3788/LOP202259.1010010 Cite this Article Set citation alerts

Shuang Li, Huafeng Li, Fan Li. Fine-Grained Cross-Modality Person Re-Identification Based on Mutual Prediction Learning[J]. Laser & Optoelectronics Progress, 2022, 59(10): 1010010 Copy Citation Text

show less

Fig. 1. Diagram of fine-grained cross-modality network for mutual prediction learning

Download full size

Fig. 2. Cross-modality identity mutual prediction learning

Download full size

Fig. 3. Effect analysis on hyperparameters

α

and

β

. (a)(b) Effect analysis of

α

; (c)(d) effect analysis of

β

Download full size

Camera	Location	（In/Out）door	Camera type	Device
1	room1	indoor	rgb	Kinect V1
2	room2	indoor	rgb	Kinect V1
3	room2	indoor	ir	-
4	gate	outdoor	rgb	-
5	garden	outdoor	rgb	-
6	passage	outdoor	ir	-

Table 1. SYSU-MM01 dataset collection environment and collection equipment

Method	Visible-thermal				Thermal-visible
Method	r=1	r=10	r=20	mAP	r=1	r=10	r=20	mAP
Zero-Pad^［24］	17.75	34.21	44.35	18.90	16.63	34.68	44.25	17.82
HCML^［30］	24.44	47.53	56.78	20.80	21.70	45.02	55.58	22.24
HSME^［19］	50.85	73.36	81.66	47.00	50.15	72.40	81.07	46.16
D2RL^［22］	43.40	66.10	76.30	44.10	‒	‒	‒	‒
MAC^［31］	36.43	62.36	71.63	37.03	36.20	61.68	70.99	39.23
AliGAN^［21］	57.90	‒	‒	53.60	56.30	‒	‒	53.40
DFE^［32］	70.13	86.32	91.96	69.14	‒	‒	‒	‒
eBDTR^［15］	34.62	58.96	68.72	33.46	34.21	58.74	68.64	32.49
MSR^［36］	48.43	70.32	79.95	48.67	‒	‒	‒	‒
JSIA^［37］	48.50	‒	‒	49.30	48.10	‒	‒	48.90
EDFL^［16］	52.58	72.10	81.47	52.98	51.89	72.09	81.04	52.13
XIV^［34］	62.21	83.13	91.72	60.18	‒	‒	‒	‒
CDP^［33］	65.00	83.50	89.60	62.70	65.3	84.5	91.0	62.1
expAT^［35］	66.48	‒	‒	67.31	67.45	‒	‒	66.51
CMSP^［14］	65.07	83.71	‒	64.50	‒	‒	‒	‒
Hi-CMD^［23］	70.93	86.39	‒	66.04	‒	‒	‒	‒
HAT^［40］	71.83	87.16	92.16	67.56	70.02	86.45	91.61	66.30
cm-SSFT^［39］	72.30	‒	‒	72.90	71.00	‒	‒	71.70
AGW^［26］	70.05	‒	‒	66.37	‒	‒	‒	‒
Ours	87.64	95.61	97.6	78.45	84.82	94.64	97.03	76.33

Table 2. Comparative experiments on RegDB dataset

Method	All-search				Indoor-search
Method	r=1	r=10	r=20	mAP	r=1	r=10	r=20	mAP
Zero-Pad^［27］	14.80	54.12	71.33	15.95	20.58	68.38	85.79	26.92
cmGAN^［29］	26.97	67.51	80.56	27.80	31.63	77.23	89.18	42.19
HCML^［30］	14.32	53.16	69.17	16.16	24.52	73.25	86.73	30.08
HSME^［19］	20.68	62.74	77.95	23.12	‒	‒	‒	‒
D2RL^［22］	28.90	70.60	82.40	29.20	‒	‒	‒	‒
MAC^［31］	33.26	79.04	90.09	36.22	36.43	62.36	71.63	37.03
AliGAN^［21］	42.40	85.00	93.70	40.70	45.90	87.60	94.40	54.30
HPILN^［17］	41.36	84.78	94.51	42.95	45.77	91.82	98.46	56.52
DFE^［32］	48.71	88.86	95.27	48.59	52.25	89.86	95.85	59.68
Hi-CMD^［23］	34.94	77.58	‒	35.94	‒	‒	‒	‒
EDFL^［16］	36.94	85.42	93.22	40.77	‒	‒	‒	‒
CDP^［33］	38.00	82.30	91.70	38.40	‒	‒	‒	‒
expAT^［35］	38.57	76.64	86.39	38.61	‒	‒	‒	‒
XIV^［34］	49.92	89.79	95.96	50.73	‒	‒	‒	‒
eBDTR^［15］	27.82	67.34	81.34	28.42	32.46	77.42	89.62	42.46
MSR^［36］	37.35	83.40	93.34	38.11	39.64	89.29	97.66	50.88
JSIA^［37］	38.10	80.70	89.90	36.90	43.80	86.20	94.20	52.90
CMSP^［14］	43.56	86.25	‒	44.98	48.62	89.50	‒	57.50
DFLA^［38］	47.14	87.93	94.45	47.08	48.03	88.13	95.14	56.84
AGW^［26］	47.50	‒	‒	47.65	54.17	‒	‒	62.97
Ours	56.48	88.25	94.43	51.91	59.69	92.80	97.55	65.86

Table 3. Comparative experiments on SYSU-MM01 dataset

Ablation study setting	SYSU-MM01（all-search）				RegDB（visible-thermal）
Ablation study setting	r=1	r=10	r=20	mAP	r=1	r=10	r=20	mAP
Baseline	48.17	82.2	89.32	45.39	67.62	85.63	91.41	62.74
Baseline+FGFL	51.22	85.59	93.40	49.43	73.69	86.46	91.21	65.81
Baseline+FGFL+CMIMPL	56.48	88.25	94.43	51.91	89.37	95.44	96.89	80.30

Table 4. Ablation experiment

Shuang Li, Huafeng Li, Fan Li. Fine-Grained Cross-Modality Person Re-Identification Based on Mutual Prediction Learning[J]. Laser & Optoelectronics Progress, 2022, 59(10): 1010010

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information