克隆/下载
贡献代码
同步代码
取消
提示: 由于 Git 不支持空文件夾,创建文件夹后会生成空的 .keep 文件
Loading...
README
MIT

Generalizing Person Re-identification

Im``plementation of ECCV2020 paper Generalizing Person Re-Identification by Camera-Aware Instance Learning and Cross-Domain Mixup.

Dependencies

Preparation

Download and extract Market-1501, DukeMTMC-reID, CUHK03 and MSMT17. Replace the root paths of corresponding datasets in the config file configs/default/dataset.py.

Train

bash train.sh GPU_ID_0,GPU_ID_1 PATH_TO_YOUR_YAML_FILE

Our code is validated under 2-GPUs setting. GPU_ID_0 and GPU_ID_1 are the indices of the selected GPUs. PATH_TO_YOUR_YAML_FILE is the path to your config yaml file. We also offer the template of config file configs/duke2market.yml, configs/market2duke.yml, configs/single_domain.yml. You can optionally adjust the hyper-parameters in the config yaml file. All of our experiments are conducted under the mix-precision training to reduce the burden of GPU memory, i.e, we set the flag fp16=true.

During the training, the checkpoint files and logs will be saved in ./checkpoints and ./logs directories, respectively.

Test

In our code, the model is evaluated on the target domain at intervals automatically. You can also evaluate the trained model manually by running:

python3 eval.py GPU_ID PATH_TO_CHECKPOINT_FILE [--dataset DATASET]

PATH_TO_CHECKPOINT_FILE is the path to the checkpoint file of the trained model. DATASET is the name of the target dataset. Its value can be {market,duke,cuhk,msmt}. As an intermediate product, the feature matrix of target images is stored in the ./features directory.

Citation

@inproceedings{luo2020generalizing,
  title={Generalizing Person Re-Identification by Camera-Aware Invariance Learning and Cross-Domain Mixup},
  author={Luo, Chuanchen and Song, Chunfeng and Zhang, Zhaoxiang},
  booktitle={European Conference on Computer Vision},
  year={2020}
}
MIT License Copyright (c) 2020 Chuanchen Luo Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

简介

原始代码 展开 收起
Python 等 6 种语言
MIT
取消

发行版

暂无发行版

贡献者

全部

近期动态

不能加载更多了
马建仓 AI 助手
尝试更多
代码解读
代码找茬
代码优化