Skip to content

Latest commit

 

History

History
32 lines (16 loc) · 1.13 KB

File metadata and controls

32 lines (16 loc) · 1.13 KB

Pytorch Zero Shot Person Re-identification via Cross View Consistency

Implementation of the paper:
Zero-Shot Person Re-identification via Cross-View Consistency
Zheng Wang, Ruimin Hu, Chao Liang, Yi Yu, Junjun Jiang, Mang Ye, Jun Chen, and Qingming Leng

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7346474

[Changes] Rather chosing from the training samples: I'm using it for cross data evaluation, i.e. the training data is replaced by the prototypes from the original dataset (called centers) and query and gallery is taken via another dataset. This is a type of cross-data evaluation.

eq2

eq2

eq3

eq3

eq4

eq4

eq5

eq5