Harmonious Attention Network for Person Re-Identification

CVPR 2018 Wei LiXiatian ZhuShaogang Gong

Existing person re-identification (re-id) methods either assume the availability of well-aligned person bounding box images as model input or rely on constrained attention selection mechanisms to calibrate misaligned images. They are therefore sub-optimal for re-id matching in arbitrarily aligned person images potentially with large human pose variations and unconstrained auto-detection errors... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Person Re-Identification CUHK03 HA-CNN MAP 38.6 # 8
Rank-1 41.7 # 9
Person Re-Identification CUHK03 detected HA-CNN (CVPR'18) MAP 38.6 # 7
Rank-1 41.7 # 7
Person Re-Identification CUHK03 labeled HA-CNN (CVPR'18) MAP 41.0 # 6
Rank-1 44.4 # 6