CLEVR_HYP: A Challenge Dataset and Baselines for Visual Question Answering with Hypothetical Actions over Images (NAACL 2021 Long Paper)
https://www.aclweb.org/anthology/2021.naacl-main.289.pdf
Download Images from the following link, keep the images corresponding to Validation Set (of original CLEVR dataset) https://dl.fbaipublicfiles.com/clevr/CLEVR_v1.0.zip
Download Action Texts and Questions (includes Train, Val, Test, 2HopAT and 2HopQH partitions) https://drive.google.com/drive/folders/1mVz9-Z548LAuLGg49_zzT42ZGKs1OT7V?usp=share_link
(1) Dataset-gen: Follow the instructions in Readme file under this directory if you would like to generate your own data in CLEVR_HYP style
(2) Baseline Model 1: Follow the instructions in Readme file under this directory if you want to evaluate performance on CLEVR_HYP using a Reading Comprehension Model (roberta-race)
(3) Baseline Model 2: Follow the instructions in Readme file under this directory if you want to evaluate performance on CLEVR_HYP using a Visual Question Answering Model (lxmert-clevr)
(4) Baseline Model 3: Follow the instructions in Readme file under this directory if you want to evaluate performance on CLEVR_HYP using the Text Editing Image Model (text-editing-image-bl)
(5) Baseline Model 4: Follow the instructions in Readme file under this directory if you want to evaluate performance on CLEVR_HYP using a Reading Comprehension Model (scene-editing-bl)