In-the-wild QA (WildQA) data and code

This repo contains the data and PyTorch code that accompanies our COLING 2022 paper:

WildQA: In-the-Wild Video Question Answering

Santiago Castro+, Naihao Deng+, Pingxuan Huang+, Mihai G. Burzo, and Rada Mihalcea

(+ equal contribution)

You can see more information at the WildQA website.

Setup

With Conda installed, run:

conda env create
conda activate wildqa

Set export WANDB_MODE=offline if you don't want to use Weights & Biases in your run.

Data

Checkout the folder src/example_data/wildQA-data/.

Run the code

For the methods presented in the paper, check out the Bash scripts under src/.

Citation

@inproceedings{castro-etal-2022-in-the-wild,
    title = "In-the-Wild Video Question Answering",
    author = "Castro, Santiago  and
      Deng, Naihao  and
      Huang, Pingxuan  and
      Burzo, Mihai G.  and
      Mihalcea, Rada",
    booktitle = "COLING",
    month = oct,
    year = "2022",
    address = "Gyeongju, Republic of Korea",
    publisher = "International Committee on Computational Linguistics",
    url = "https://aclanthology.org/2022.coling-1.496",
    pages = "5613--5635",
}