【2020-04-06更新】跨模态行人再识别汇总

Methods	RegDB		SYSU-MM01
	Rank-1(%)	mAP(%)	Rank-1(%)	mAP(%)
Deep Zero-Padding(ICCV2017)	–	–	14.80	15.95
HCML(AAAI2018)	24.44	20.80	–	–
cmGAN(IJCAI2018)	–	–	26.97	27.80
BDTR(IJCAI2018)	33.47	31.83	17.01	19.66
D2RL(CVPR2019)	43.40	44.10	28.90	29.20
HSME(AAAI2019)	50.85	47.00	20.68	23.12
IPVT-1 and MSR(Access2019)	58.76	47.85	23.18	22.49
EDFL(ArXiv2019)	52.58	52.98	36.94	40.77
AlignGAN(ICCV2019)	56.30	53.40	42.40	40.70
HPILN(IET IP2019)	–	–	41.36	42.95
TSLFN+HC(ArXiv2019)	83.00	72.00	56.96	54.95
DSCSN+CCN(ArXiv2019)	60.80	60.00	35.10	37.40
JSIA(AAAI2020)	–	–	38.10	36.90
Hi-CMD(CVPR2020)	70.44	65.93	34.94	35.94
cm-SSFT(CVPR2020)	72.30	72.90	61.60	63.20
BDTR(IJCAI2018新版本AWG)	70.05	66.37	47.50	47.65

1. 2017-ICCV-RGB-Infrared Cross-Modality Person Re-Identification

	Deep Zero-Padding

2. 2018-AAAI-Hierarchical Discriminative Learning for Visible Thermal Person Re-Identification

    Hierarchical Cross-modality Metric Learning  (HCML)

3. 2018-IJCAI-Cross-Modality Person Re-Identification with Generative Adversarial Training

    cross-modality Generative Adversarial Network (cmGAN)

4. 2018-IJCAI-Visible thermal person re-identification via dual-constrained top-ranking

    Bi-directional Dual-constrained Top-Ranking (BDTR)

5. 2019-CVPR-Learning to Reduce Dual-level Discrepancy for Infrared-Visible Person Re-identification

    Dual-level Discrepancy Reduction Learning（D2RL）

6. 2019-AAAI-HSME: Hypersphere Manifold Embedding for Visible Thermal Person Re-Identification

    HyperSphere Manifold Embedding (HSME)

7. 2019-IEEE Access-Person Re-Identification Between Visible and Thermal Camera Images Based on Deep Residual CNN Using Single Input

    (IPVT-1 and MSR)    

8. 2019-ArXiv-Enhancing the Discriminative Feature Learning for Visible-Thermal Cross-Modality Person Re-Identification

    Enhancing Discriminative Feature Learning (EDFL)

9. 2019-ICCV-RGB-Infrared Cross-Modality Person Re-Identification via Joint Pixel and Feature Alignment

	Alignment Generative Adversarial Network (AlignGAN)

10.2019-IET IP-HPILN: A feature learning framework for cross-modality person re-identification

	Hard Pentaplet and Identity Loss Network (HPILN)

11. 2019-ArXiv-Hetero-Center Loss for Cross-Modality Person Re-Identification

	Two-Stream Local Feature Network (TSLFN)+Hetero-Center loss (HC)

12.2019-ArXiv-Attend to the Difference: Cross-Modality Person Re-identification via Contrastive Correlation

	Dual-path Spatial-structure-preserving Common Space Network (DSCSN) + Contrastive Correlation Network (CCN)

13.2020-CVPR-Hi-CMD: Hierarchical Cross-Modality Disentanglement for Visible-Infrared Person Re-Identification

	Hi-CMD

14.2020-CVPR-Cross-modality Person re-identification with Shared-Specific Feature Transfer

	cm-SSFT

15.2020-AAAI-Cross-Modality Paired-Images Generation for RGB-InfraredPerson Re-Identification

	JSIA

数据集

RegDB Dataset 【1】

SYSU-MM01 Dataset 【2】

【1】2017-Sensor-Person Recognition System Based on a Combination of Body Images from Visible Light and Thermal Cameras
【2】2017-ICCV-RGB-Infrared Cross-Modality Person Re-Identification

公开代码

叶茫博士的代码：
https://github.com/mangye16/Cross-Modal-Re-ID-baseline
新版代码：
https://github.com/liuliu408/Cross-Modal-Re-ID-baseline_2020

TSLFN+HC(ArXiv2019)代码：
https://codeload.github.com/98zyx/Hetero-center-loss-for-cross-modality-person-re-id/zip/master

HiCMD(CVPR2020)代码：
https://github.com/bismex/HiCMD

JSIA(AAAI2020)代码：
https://github.com/wangguanan/JSIA-ReID