Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification

ICLR 2020 Yixiao GeDapeng ChenHongsheng Li

Person re-identification (re-ID) aims at identifying the same persons' images across different cameras. However, domain diversities between different datasets pose an evident challenge for adapting the re-ID model trained on one dataset to another one... (read more)

PDF Abstract ICLR 2020 PDF ICLR 2020 Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Unsupervised Person Re-Identification DukeMTMC-reID->Market-1501 MMT-ResNet50 mAP 71.2 # 1
Top-1 (%) 87.7 # 1
Unsupervised Person Re-Identification DukeMTMC-reID->MSMT17 MMT-ResNet50 mAP 23.5 # 2
Top-1 (%) 50.0 # 1
Unsupervised Domain Adaptation Duke to Market MMT mAP 71.2 # 3
rank-1 87.7 # 3
rank-5 94.9 # 3
rank-10 96.9 # 3
Unsupervised Domain Adaptation Duke to MSMT MMT mAP 23.3 # 4
rank-1 50.1 # 4
rank-5 63.9 # 4
rank-10 69.8 # 4
Unsupervised Person Re-Identification Market-1501->DukeMTMC-reID MMT-ResNet50 mAP 65.1 # 1
Rank-1 78.0 # 1
Rank-10 88.8 # 1
Rank-5 92.5 # 1
Unsupervised Person Re-Identification Market-1501->MSMT17 MMT-ResNet50 mAP 22.9 # 1
Top-1 (%) 49.2 # 1
Unsupervised Domain Adaptation Market to Duke MMT mAP 65.1 # 3
rank-1 78.0 # 3
rank-5 88.8 # 3
rank-10 92.5 # 2
Unsupervised Domain Adaptation Market to MSMT MMT mAP 22.9 # 4
rank-1 49.2 # 4
rank-5 63.1 # 3
rank-10 68.8 # 3

Methods used in the Paper