stevetsui/rbonn

Recurrent Bilinear Optimization for Binary Neural Networks (RBONN)

Pytorch implementation of our paper "Recurrent Bilinear Optimization for Binary Neural Networks" accepted by ECCV2022 as oral presentation.

Tips

Any problem, please contact the first author (Email: [email protected]).

Our code is heavily borrowed from ReActNet (https://github.com/liuzechun/ReActNet).

Dependencies

• Python 3.8
• Pytorch 1.7.1
• Torchvision 0.8.2

RBONN with two-stage tranining

We test our RBONN using the same ResNet-18 structure and training setttings as ReActNet, and obtain 66.7% top-1 accuracy.

Methods	Top-1 acc	Top-5 acc	Quantized model link	Log
ReActNet	65.9	-	Model	-
ReCU	66.4	86.5	Model	-
RBONN	66.7	87.0	Model	Log

To verify the performance of our quantized models with ReActNet-like structure on ImageNet, please do as the following steps:

1. Finish the first stage training using ReActNet.
2. Use the following command:

cd 2_step2_rbonn 
bash run.sh

If you find this work useful in your research, please consider to cite:

@inproceedings{xu2022recurrent,
  title={Recurrent bilinear optimization for binary neural networks},
  author={Xu, Sheng and Li, Yanjing and Wang, Tiancheng and Ma, Teli and Zhang, Baochang and Gao, Peng and Qiao, Yu and L{\"u}, Jinhu and Guo, Guodong},
  booktitle={European Conference on Computer Vision},
  pages={19--35},
  year={2022},
  organization={Springer}
}

rbonn's Issues

BNN Deployment

Hi, thanks for the great work. Is it convenient for you to open source the deployment code of BNN, which will greatly promote the BNN research community.