Omni-Scale Feature Learning for Person Re-Identification

ICCV 2019 Kaiyang ZhouYongxin YangAndrea CavallaroTao Xiang

As an instance-level recognition problem, person re-identification (ReID) relies on discriminative features, which not only capture different spatial scales but also encapsulate an arbitrary combination of multiple scales. We call features of both homogeneous and heterogeneous scales omni-scale features... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Person Re-Identification CUHK03 OSNet MAP 67.8 # 2
Person Re-Identification DukeMTMC-reID OSNet Rank-1 88.6 # 8
Person Re-Identification DukeMTMC-reID OSNet MAP 73.5 # 12
Person Re-Identification Market-1501 OSNet Rank-1 94.8 # 7
Person Re-Identification Market-1501 OSNet MAP 84.9 # 11
Person Re-Identification MSMT17 OSNet Rank-1 78.7 # 2
Person Re-Identification MSMT17 OSNet mAP 52.9 # 2