Models and Codes for the paper: Question Relevance in VQA: Identifying Non-Visual And False-Premise Questions, https://arxiv.org/pdf/1606.06622v2.pdf
- For Question-Caption or Question-Question Match Models, look at the
questionCaptionMatchModels/
folder. - For Visual vs. Non Visual Question Identification Models, look at the
visualNonVisualModels/
folder. - For VNQ and VTFQ data, look at the
data/
folder. Note that the models in the above two folders do not load data directly from thisdata/
folder. This is just the raw data.
Look at the README's inside the individual folders for more information.
Please email [email protected] if you have any questions/concerns.