Welcome to MMCultureQA 2027
MMCultureQA is a SemEval 2027 shared task on visual question answering in multiple languages. A system is given an image and a question about it, and has to write a short, open-ended answer. The question can be spoken or typed, and it comes in multiple languages.
What makes the task hard is cultural grounding. Many questions are about the food, places, customs, and objects, so the right answer often depends on local knowledge rather than on what is plainly visible in the image.
Tasks
The shared task has two tasks, each offered as a separate track per language variety. You may enter either task or both.
- Task 1: Spoken Visual QA generates an answer to a spoken (audio) question about an image.
- Task 2: Textual Visual QA answers the same question presented as text.
See the Tasks page for the full definition, and the OASIS page for the dataset the shared task is built on.
Important Dates
All dates are tentative and will be confirmed at the data release.
- 15 July 2026: Sample data release Find here
- 1 September 2026: Training data release Next
- 10 January 2027: Evaluation phase begins
- 31 January 2027: Evaluation phase ends
- February 2027: System description papers due
- March 2027: Notification to authors
- April 2027: Camera-ready papers due
- Summer 2027: SemEval 2027 workshop
Recent Updates
- 14 June 2026: Teams can now sign up to take part in MMCultureQA. Register here.
- 13 June 2026: Explore the task definition, the two tasks, the OASIS dataset, the evaluation plan, important dates, and the organizers.
- 13 June 2026: A sample set will be released first so teams can preview the format. Training data follows on 1 September 2026.
Contact
Registration is open: sign up using the registration form. For full details on taking part, see the Participate page. A participant mailing list is being set up and will be linked here, and for other questions you can contact the organizers.