AlignedReID: Surpassing Human-Level Performance in Person Re-Identification

22 Nov 2017  ·  Xuan Zhang, Hao Luo, Xing Fan, Weilai Xiang, Yixiao Sun, Qiqi Xiao, Wei Jiang, Chi Zhang, Jian Sun ·

In this paper, we propose a novel method called AlignedReID that extracts a global feature which is jointly learned with local features. Global feature learning benefits greatly from local feature learning, which performs an alignment/matching by calculating the shortest path between two sets of local features, without requiring extra supervision. After the joint learning, we only keep the global feature to compute the similarities between images. Our method achieves rank-1 accuracy of 94.4% on Market1501 and 97.8% on CUHK03, outperforming state-of-the-art methods by a large margin. We also evaluate human-level performance and demonstrate that our method is the first to surpass human-level performance on Market1501 and CUHK03, two widely used Person ReID datasets.

PDF Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Person Re-Identification CUHK03 AlignedReID (RK) Rank-1 97.8 # 2
Rank-5 99.6 # 1
Rank-10 99.8 # 1
Person Re-Identification CUHK03-C Aligned++ Rank-1 7.99 # 5
mAP 4.87 # 4
mINP 0.56 # 4
Person Re-Identification CUHK-SYSU AlignedReID MAP 94.4 # 1
Rank-1 95.7 # 1
Person Re-Identification Market-1501 AlignedReID (RK) Rank-1 94.4 # 66
mAP 90.7 # 34
Person Re-Identification Market-1501-C Aligned++ Rank-1 31.00 # 14
mAP 10.95 # 9
mINP 0.32 # 10


No methods listed for this paper. Add relevant methods here