README.md

March 30, 2021 · View on GitHub

Deformable ConvNets v2

骨架网络网络类型卷积每张GPU图片个数学习率策略推理时间(fps)Box APMask AP下载配置文件
ResNet50-FPNFasterc3-c511x-42.1-下载链接配置文件
ResNet50-vd-FPNFasterc3-c511x-42.7-下载链接配置文件
ResNet50-vd-FPNFasterc3-c512x-43.7-下载链接配置文件
ResNet101-vd-FPNFasterc3-c511x-45.1-下载链接配置文件
ResNeXt101-vd-FPNFasterc3-c511x-46.5-下载链接配置文件
ResNet50-FPNMaskc3-c511x-42.738.4下载链接配置文件
ResNet50-vd-FPNMaskc3-c512x-44.639.8下载链接配置文件
ResNet101-vd-FPNMaskc3-c511x-45.640.6下载链接配置文件
ResNeXt101-vd-FPNMaskc3-c511x-47.342.0下载链接配置文件
ResNet50-FPNCascade Fasterc3-c511x-42.1-下载链接配置文件
ResNeXt101-vd-FPNCascade Fasterc3-c511x-48.8-下载链接配置文件

注意事项:

  • Deformable卷积网络v2(dcn_v2)参考自论文Deformable ConvNets v2.
  • c3-c5意思是在resnet模块的3到5阶段增加dcn.

Citations

@inproceedings{dai2017deformable,
  title={Deformable Convolutional Networks},
  author={Dai, Jifeng and Qi, Haozhi and Xiong, Yuwen and Li, Yi and Zhang, Guodong and Hu, Han and Wei, Yichen},
  booktitle={Proceedings of the IEEE international conference on computer vision},
  year={2017}
}
@article{zhu2018deformable,
  title={Deformable ConvNets v2: More Deformable, Better Results},
  author={Zhu, Xizhou and Hu, Han and Lin, Stephen and Dai, Jifeng},
  journal={arXiv preprint arXiv:1811.11168},
  year={2018}
}