Domain Adaptive Person Re-Identification via Coupling Optimization

6 Nov 2020  ·  Xiaobin Liu, Shiliang Zhang ·

Domain adaptive person Re-Identification (ReID) is challenging owing to the domain gap and shortage of annotations on target scenarios. To handle those two challenges, this paper proposes a coupling optimization method including the Domain-Invariant Mapping (DIM) method and the Global-Local distance Optimization (GLO), respectively. Different from previous methods that transfer knowledge in two stages, the DIM achieves a more efficient one-stage knowledge transfer by mapping images in labeled and unlabeled datasets to a shared feature space. GLO is designed to train the ReID model with unsupervised setting on the target domain. Instead of relying on existing optimization strategies designed for supervised training, GLO involves more images in distance optimization, and achieves better robustness to noisy label prediction. GLO also integrates distance optimizations in both the global dataset and local training batch, thus exhibits better training efficiency. Extensive experiments on three large-scale datasets, i.e., Market-1501, DukeMTMC-reID, and MSMT17, show that our coupling optimization outperforms state-of-the-art methods by a large margin. Our method also works well in unsupervised training, and even outperforms several recent domain adaptive methods.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Unsupervised Person Re-Identification DukeMTMC-reID->Market-1501 DIM+GLO mAP 65.1 # 3
Rank-1 88.3 # 1
Unsupervised Person Re-Identification DukeMTMC-reID->MSMT17 DIM+GLO mAP 24.4 # 1
Rank-1 56.5 # 1
Unsupervised Person Re-Identification Market-1501->DukeMTMC-reID DIM+GLO mAP 58.3 # 2
Rank-1 76.2 # 2
Unsupervised Person Re-Identification Market-1501->MSMT17 DIM+GLO mAP 20.7 # 2
Rank-1 49.7 # 1


No methods listed for this paper. Add relevant methods here