Code for creating dataset and running models described in the paper "A dataset and models for understanding video data through fill-in-the-blank question-answering", which will appear at CVPR2017.

Paper: https://arxiv.org/abs/1611.07810

MovieFIB was used in the Fill-in-the-Blank track of the LSMDC2016 (Large Scale Movie Description and Understanding Challenge) at ECCV2016, and will be used in LSMDC2017 at ICCV2017; details to be announced soon.

The dataset is available for download here: https://sites.google.com/site/describingmovies/lsmdc-2016/download