About the reference dataset in Phase 2

In the paper, " The YFCC100M dataset is publicly available, but we will re-host the chosen subset." is argued. Does that mean the reference dataset in Phase 2 will also be different?
However, in the problem description set on this website, NO new reference dataset is pointed.

I want to know whether we will have a new reference dataset. If the answer is Yes, it is very difficult to download these one million data in two days in some countries, like China. Is there a solution for this? Thanks.

1 Like

Hi wenhaowang. The reference set in Phase 2 will be the same. Only the query set will be different.

(The sentence you quoted is just referring to the fact that we are hosting the competition data ourselves rather than having participants use the publicly available YFCC100M data.)

Hope that helps!

1 Like

It helps a lot! Thanks very much!

1 Like