News - ICDAR 2019 Robust Reading Challenge on Multi-lingual scene text detection and recognition


QA Types made public

We are glad to announce that we have make public the question types for Single-Page DocVQA and InfographicsVQA datasets. However, we won't release it for the test sets, as we believe that this should not be used as any kind of help at the time of answering the questions, but to better interpret the method's results. In addition, the public evaluation tool in the RRC platform already provides this information.

Therefore, we have released the question types only for training and validation set of the SP-DocVQA, and validation set of InfographicsVQA. You can find the updated links in the downloads section.

We hope this helps you in your research,

The DocVQA team

Important Dates

15 Feb to 2 May

Manifestation of interest by participants opens

Asking/Answering questions about the details of the competition

1 Mar

Competition formal announcement

15 Mar

Website fully ready

Registration of participants continues

Evaluation protocol, file formats etc. available

15 Mar to 2 May

Train set available - training period - MLT challenge in progress -Participants evaluate their methods on the training/validation sets - Prepare for submission

Registration is still open

2 May

Registration closes for this MLT challenge for ICDAR-2019

2 May to 3 June

Test set available

27 May

Deadline for submitting: 1) participant information (names and affiliation), 2) methods description, 3) initial (or final) results

3 June

Deadline for submission of the final results by participants

20 - 25 Sept

Announcement of results at ICDAR2019

1 Oct

The public release of the full dataset