Home

Awesome

Incremental Learning Techniques for Semantic Segmentation

This is the code of the paper: Michieli U., Zanuttigh P., "Incremental Learning Techniques for Semantic Segmentation", Proceedings of the International Conference on Computer Vision (ICCV), Workshop on Transferring and Adapting Source Knowledge in Computer Vision (TASK-CV), Seoul (South Korea), 2 November 2019.

If you find the code or the paper useful for your research please cite our work as:

@inproceedings{michieli2019,
  title={Incremental Learning Techniques for Semantic Segmentation},
  author={Michieli, Umberto and Zanuttigh, Pietro},
  booktitle={International Conference on Computer Vision (ICCV), Workshop on Transferring and Adapting Source Knowledge in Computer Vision (TASK-CV)},
  year={2019}
}

The PDF of the paper can be found at: https://arxiv.org/abs/1907.13372

The webpage of the paper is: https://lttm.dei.unipd.it/paper_data/IL/

architecture image

Setup

Training

I suggest to start from the folder codes/LD1/ where you can run the simulations for the incremental scenario in which only the last class is added and only the LD1 loss is present.
The procedure is discussed in detail at codes/LD1/README.txt.
NB: for the other folders no README.txt exists since they are straightforward extensions of the code contained in the LD1/ folder

Software & Hardware

Contact Information

I can provide checkpoints, datasets or other material upon specific request.

Umberto Michieli
Ph.D. Student
University of Padova
Department of Information Engineering (DEI)
Email: umberto.michieli@dei.unipd.it
Personal Website: https://umbertomichieli.github.io/

License

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.