andeeptoor / qar-qae

Code and dataset for paper on "Question action relevance and editing for visual question answering"

qar-qae

Dataset and model-building code for "Question action relevance and editing for visual question answering" paper.

These datasets are based on the Visual Genome Dataset and the image filenames that are referenced come from that dataset.

If you find this useful, please use the following citation:

@article{toor2018question,
  title={Question action relevance and editing for visual question answering},
  author={Toor, Andeep S and Wechsler, Harry and Nappi, Michele},
  journal={Multimedia Tools and Applications},
  pages={1--15},
  year={2018},
  publisher={Springer US}
}