Lightweight Probabilistic Deep Networks

This source code accompanies the paper:

Lightweight Probabilistic Deep Networks, Jochen Gast and Stefan Roth, CVPR 2018.

Framework Requirements

This code has been built and tested with PyTorch 0.4.0 on Python 2.7. Please check PyTorch for requirements and setup instructions. Typically, the easiest way is to install a full Anaconda Python distribution.

Custom packages

After having installed Anaconda, following custom packages are required: setproctitle tqdm. They can be installed via the one-liner

$ pip install setproctitle tqdm

Pretrained models

Model	Checkpoint
FlowNetADF	Link
FlowNetProbOut	Link

Contact

If you have further questions, feel free to contact jochen.gast@visinf.tu-darmstadt.de .

Citation

If you use our code, please cite our CVPR 2018 paper:

@inproceedings{Gast:2018:LPD,
  title={Lightweight Probabilistic Deep Networks},
  author={Jochen Gast and Stefan Roth},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  year={2018}
}