CoSE: Compositional Stroke Embeddings

This repository is forked from CoSE repository for my thesis project.

Set up the workflow

Environmenmt

Configure the virtual environment by running the following bash command targetting at environment configuration bash script:

chmod +x environment_configuration.sh
source environment_configuration.sh 
conda deactivate

Pre-trained Models

The pre-trained model can be downloaded from authors' shared Google Drive.

Trained Dataset

One of the trained dataset of CoSE model is QuickDraw dataset. Note that CoSE model requires raw files.

thesis directory

Citation

@article{aksan2020cose,
  title={CoSE: Compositional Stroke Embeddings},
  author={Aksan, Emre and Deselaers, Thomas and Tagliasacchi, Andrea and Hilliges, Otmar},
  journal={Advances in Neural Information Processing Systems},
  volume={33},
  year={2020}
}

@article{gervais2020didi,
  title={The DIDI dataset: Digital Ink Diagram data},
  author={Gervais, Philippe and Deselaers, Thomas and Aksan, Emre and Hilliges, Otmar},
  journal={arXiv preprint arXiv:2002.09303},
  year={2020}
}