ABD-Net: Attentive but Diverse Person Re-Identification

ICCV 2019 Tianlong ChenShaojin DingJingyi XieYe YuanWuyang ChenYang YangZhou RenZhangyang Wang

Attention mechanism has been shown to be effective for person re-identification (Re-ID). However, the learned attentive feature embeddings which are often not naturally diverse nor uncorrelated, will compromise the retrieval performance based on the Euclidean distance... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Person Re-Identification DukeMTMC-reID ABD-Net (ResNet-50) Rank-1 89.0 # 7
Person Re-Identification DukeMTMC-reID ABD-Net (ResNet-50) MAP 78.59 # 9
Person Re-Identification DukeMTMC-reID ABD-Net (ResNet-50) Top 1 Accuracy 89.0 # 1
Person Re-Identification Market-1501 ABD-Net (ResNet-50) Rank-1 95.6 # 5
Person Re-Identification Market-1501 ABD-Net (ResNet-50) MAP 88.28 # 10
Person Re-Identification MSMT17 ABD-Net (ResNet-50) Rank-1 82.3 # 1
Person Re-Identification MSMT17 ABD-Net (ResNet-50) mAP 60.8 # 1