ICCV 2019 Workshop & Challenge on Computer Vision for Wildlife Conservation (CVWC)

Recent Updates

[2019.8.4]

Note

   We participate in the Plain Re-ID Track. Our solution uses SE-ResNet50 model as backbone which was pre-trained by ILSVRC. In addition, we design two complementary network branches to learn multiple discriminative features. We use multi-task learning strategy to supervise the model training. Finally, we fine-tune the model with triplet loss. The Re-ID results are obtained based on the fusion of the learned multiple features.

Dependencies

• python == 3.6
• torch == 0.4.1
• torchvision == 0.2.1

Solution

https://github.com/liuning-scu-cn/AmurTigerReID

Prepare Data

• Train data:

Please download from https://lilablobssc.blob.core.windows.net/cvwc2019/train/atrw_reid_train.tar.gz

• Train Annotations:

Please download from https://lilablobssc.blob.core.windows.net/cvwc2019/train/atrw_anno_reid_train.tar.gz

• Test data:

Please download from https://lilablobssc.blob.core.windows.net/cvwc2019/test/atrw_reid_test.tar.gz

• Val data:

Please download from ...

Competition Dataset

train_set   =>   ./dataload/dataloader.py
test_set    =>   ./dataload/dataloader.py

Train

In train.py, finetune_tiger_cnn5.py, and finetune_tiger_cnn8.py.

If you want to run finetune_tiger_cnn5.py, you firstly need to train tiger_cnn1 model.

If you want to run finetune_tiger_cnn8.py, you firstly need to train tiger_cnn3 model.

Test

In test.py

Model

If you want to get our all models and logs, please download from https://pan.baidu.com/s/11RslAFW9g7kS8I4IVZC_Iw, and passward code is 7iic. Then you can save it to ./model.

Plain ReID Track

Thanks

• CVWC 2019 Organizing Team
• EvalAI Team
• SiChuan University BRL

If you have any problems, please contact BRL or email to [email protected].