Learning Compatible Embeddings, ICCV 2021

by Qiang Meng, Chixiang Zhang, Xiaoqiang Xu and Feng Zhou

Paper: Arxiv

We cannot release source codes publicly for the interests of the company. This repo (modified from https://github.com/guxinqian/Simple-ReID) are the codes for the ReID experiments as well as core modules/steps. It should be not hard to implement the method in face recognition based on given codes.

@inproceedings{meng2021lce,
  title={Learning Compatible Embeddings},
  author={Meng, Qiang and Zhang, Chixiang and Xu, Xiaoqiang and Zhou, Feng},
  booktitle=ICCV,
  year=2021
}

1. Prepare the Market1501 dataset
2. Train the base model by ./scripts/train.sh.
3. Train the LCE model by .scripts/train_lce.sh.
4. Evaluate the performances by ./scripts/test.sh.