News - Out of Vocabulary Scene Text Understanding

Multipage DocVQA Dataset is public!

We have released the dataset for Multipage Document Visual Question Answering (MP-DocVQA). You can find it in the Downloads section under the Task 4 section.

We are currently working on the evaluation protocol to allow you submit your results and be ranked in the leaderboard, but currently it is still closed.

If you are interested in this task, you can find more information in the paper Hierarchical multimodal transformers for Multi-Page DocVQA.