FloodNet Challenge @ EARTHVISION 2021 - Track 2 Forum

Go back to competition Back to thread list Post in this thread

> Could we use segmentation model trained from Track1 data to help VQA?

As above.

Posted by: EVER @ April 4, 2021, 4:43 a.m.

As VQA is a separate task from semantic segmentation, we don't inspire participant to use semantic segmented model for VQA task. We inspire all to come with new models where inputs are only images and questions.

Thank you.

Posted by: binalab @ April 5, 2021, 4:11 p.m.
Post in this thread