Localized Medical VQA

<p>This dataset contains 3 sub-datasets with questions about regions for the Medical Visual Question Answering (VQA) task. Traditionally, questions are asked about the entire image. In these datasets, we ask questions about randomly generated regions in an image for fundus images as well as cataract surgery frames and surgeries performed with the DaVinci robot.</p><p>All three datasets were created using publicly available datasets.</p><p>For more information and code, visit <a href="https://github.com/sergiotasconmorales/locvqa">our GitHub page.</a></p><p>If you use this dataset, please cite our work:</p><blockquote><p>@inproceedings{tascon2023localized, &nbsp;<br>title={Localized Questions in Medical Visual Question Answering}, &nbsp;<br>author={Tascon-Morales, Sergio and M{\'a}rquez-Neila, Pablo and Sznitman, Raphael}, &nbsp;<br>booktitle={International Conference on Medical Image Computing and Computer-Assisted Intervention}, &nbsp;<br>pages={361--370}, &nbsp;<br>year={2023}, &nbsp;<br>organization={Springer} <br>} </p></blockquote>