Collaborative Refining for Person Re-Identification With Label Noise

Existing person re-identification (Re-ID) methods usually rely heavily on large-scale thoroughly annotated training data. However, label noise is unavoidable due to inaccurate person detection results or annotation errors in real scenes. It is extremely challenging to learn a robust Re-ID model with...

Ausführliche Beschreibung

Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on image processing : a publication of the IEEE Signal Processing Society. - 1992. - 31(2022) vom: 07., Seite 379-391
1. Verfasser: Ye, Mang (VerfasserIn)
Weitere Verfasser: Li, He, Du, Bo, Shen, Jianbing, Shao, Ling, Hoi, Steven C H
Format: Online-Aufsatz
Sprache:English
Veröffentlicht: 2022
Zugriff auf das übergeordnete Werk:IEEE transactions on image processing : a publication of the IEEE Signal Processing Society
Schlagworte:Journal Article
LEADER 01000naa a22002652 4500
001 NLM334114942
003 DE-627
005 20231225223031.0
007 cr uuu---uuuuu
008 231225s2022 xx |||||o 00| ||eng c
024 7 |a 10.1109/TIP.2021.3131937  |2 doi 
028 5 2 |a pubmed24n1113.xml 
035 |a (DE-627)NLM334114942 
035 |a (NLM)34874857 
040 |a DE-627  |b ger  |c DE-627  |e rakwb 
041 |a eng 
100 1 |a Ye, Mang  |e verfasserin  |4 aut 
245 1 0 |a Collaborative Refining for Person Re-Identification With Label Noise 
264 1 |c 2022 
336 |a Text  |b txt  |2 rdacontent 
337 |a ƒaComputermedien  |b c  |2 rdamedia 
338 |a ƒa Online-Ressource  |b cr  |2 rdacarrier 
500 |a Date Completed 13.12.2021 
500 |a Date Revised 14.12.2021 
500 |a published: Print-Electronic 
500 |a Citation Status MEDLINE 
520 |a Existing person re-identification (Re-ID) methods usually rely heavily on large-scale thoroughly annotated training data. However, label noise is unavoidable due to inaccurate person detection results or annotation errors in real scenes. It is extremely challenging to learn a robust Re-ID model with label noise since each identity has very limited annotated training samples. To avoid fitting to the noisy labels, we propose to learn a prefatory model using a large learning rate at the early stage with a self-label refining strategy, in which the labels and network are jointly optimized. To further enhance the robustness, we introduce an online co-refining (CORE) framework with dynamic mutual learning, where networks and label predictions are online optimized collaboratively by distilling the knowledge from other peer networks. Moreover, it also reduces the negative impact of noisy labels using a favorable selective consistency strategy. CORE has two primary advantages: it is robust to different noise types and unknown noise ratios; it can be easily trained without much additional effort on the architecture design. Extensive experiments on Re-ID and image classification demonstrate that CORE outperforms its counterparts by a large margin under both practical and simulated noise settings. Notably, it also improves the state-of-the-art unsupervised Re-ID performance under standard settings. Code is available at https://github.com/mangye16/ReID-Label-Noise 
650 4 |a Journal Article 
700 1 |a Li, He  |e verfasserin  |4 aut 
700 1 |a Du, Bo  |e verfasserin  |4 aut 
700 1 |a Shen, Jianbing  |e verfasserin  |4 aut 
700 1 |a Shao, Ling  |e verfasserin  |4 aut 
700 1 |a Hoi, Steven C H  |e verfasserin  |4 aut 
773 0 8 |i Enthalten in  |t IEEE transactions on image processing : a publication of the IEEE Signal Processing Society  |d 1992  |g 31(2022) vom: 07., Seite 379-391  |w (DE-627)NLM09821456X  |x 1941-0042  |7 nnns 
773 1 8 |g volume:31  |g year:2022  |g day:07  |g pages:379-391 
856 4 0 |u http://dx.doi.org/10.1109/TIP.2021.3131937  |3 Volltext 
912 |a GBV_USEFLAG_A 
912 |a SYSFLAG_A 
912 |a GBV_NLM 
912 |a GBV_ILN_350 
951 |a AR 
952 |d 31  |j 2022  |b 07  |h 379-391