Rethinking the Distribution Gap of Person Re-identification with Camera-based Batch Normalization

The fundamental difficulty in person re-identification (ReID) lies in learning the correspondence among individual cameras. It strongly demands costly inter-camera annotations, yet the trained models are not guaranteed to transfer well to previously unseen cameras... These problems significantly limit the application of ReID. This paper rethinks the working mechanism of conventional ReID approaches and puts forward a new solution. With an effective operator named Camera-based Batch Normalization (CBN), we force the image data of all cameras to fall onto the same subspace, so that the distribution gap between any camera pair is largely shrunk. This alignment brings two benefits. First, the trained model enjoys better abilities to generalize across scenarios with unseen cameras as well as transfer across multiple training sets. Second, we can rely on intra-camera annotations, which have been undervalued before due to the lack of cross-camera information, to achieve competitive ReID performance. Experiments on a wide range of ReID tasks demonstrate the effectiveness of our approach. The code is available at https://github.com/automan000/Camera-based-Person-ReID. read more

PDF Abstract ECCV 2020 PDF ECCV 2020 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Person Re-Identification DukeMTMC-reID CBN+BoT* Rank-1 84.8 # 46
Rank-5 92.5 # 6
Rank-10 95.2 # 4
mAP 70.1 # 47
Person Re-Identification DukeMTMC-reID CBN Rank-1 82.5 # 49
Rank-5 91.7 # 7
Rank-10 94.1 # 6
mAP 67.3 # 51
Unsupervised Domain Adaptation Duke to Market CBN+ECN mAP 52 # 13
rank-1 81.7 # 9
rank-5 91.9 # 8
rank-10 94.7 # 8
Person Re-Identification Market-1501 CBN+BoT* Rank-1 94.3 # 45
Rank-5 97.9 # 7
Rank-10 98.7 # 5
mAP 83.6 # 52
Person Re-Identification Market-1501 CBN Rank-1 91.3 # 56
Rank-5 97.1 # 9
Rank-10 98.4 # 6
mAP 77.3 # 63
Unsupervised Domain Adaptation Market to Duke CBN+ECN mAP 44.9 # 14
rank-1 68 # 13
rank-5 80 # 10
rank-10 83.9 # 8
Person Re-Identification MSMT17 CBN Rank-1 72.8 # 20
mAP 42.9 # 20

Methods