Unsupervised Pre-training (ResNet101+MGN)

Reported on 6 benchmarks across 1 task · 1 paper · 1 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision6 results

Person Re-IdentificationonMSMT17
Rank-1· uses extra data· 2020-12-07
86.6
best: 91.7 (SOLIDER (with re-ranking))
SOTA
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753
Person Re-IdentificationonMSMT17
mAP· uses extra data· 2020-12-07
68.8
best: 86.7 (CLIP-ReID (with re-ranking))
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753
Person Re-IdentificationonMarket-1501
Rank-1· 2020-12-07
97
best: 98 (st-ReID(RE, RK))
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753
Person Re-IdentificationonMarket-1501
mAP· 2020-12-07
92
best: 96.21 (Unsupervised Pre-training (ResNet101+RK))
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753
Person Re-IdentificationonDukeMTMC-reID
Rank-1· 2020-12-07
91.9
best: 95.6 (CTL Model (ResNet50, 256x128))
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753
Person Re-IdentificationonDukeMTMC-reID
mAP· 2020-12-07
84.1
best: 97.1 (DenseIL)
Unsupervised Pre-training for Person Re-identification arXiv:2012.03753