The pytorch implementation of Position-Aware Relation Networks (PARN), which is proposed in Position-Aware Relation Networks for Few-Shot Learning.

• Build environments of Deformable Convolutional Networks (DCN)
  • Download the DCN code, and follow the guideline to build environments of DCN. If you fail to complie the above code, you can try this code.
• Download the Mini-Imagenet dataset, and put it in the datas/miniImagenet folder.
• Run pip install -r requirements.txt
• Run python main.py

@article{wu2019parn,
    title={PARN: Position-Aware Relation Networks for Few-Shot Learning},
    author={Wu, Ziyang and Li, Yuwei and Guo, Lihua and Jia, Kui},
    journal={ICCV},
    year={2019}
}