liuyyy111

你用SCAN的feature train出来了吗

> > 请问VLP 里有全局的feature下载 和regional的feature下载，是应该下载哪个？ > > # data > > wget -O COCO00 "https://onedrive.live.com/download?cid=E5364FD183A1F5BB&resid=E5364FD183A1F5BB%212019&authkey=ACn4bwZ0nmZ0nik" > > wget -O COCO01 "https://onedrive.live.com/download?cid=E5364FD183A1F5BB&resid=E5364FD183A1F5BB%212018&authkey=AHoTGG-7-6kwoAY" > > wget -O flickr30k.tar.gz "https://onedrive.live.com/download?cid=E5364FD183A1F5BB&resid=E5364FD183A1F5BB%212015&authkey=AFZ2iehPM8HREeA" > > wget...

in my case, I got an unbelievable result after using the t2i rerank, did you have same problem?

thank you for your prompt reply. The "unbelievable results" means that I got the result of t2i is almost slightly better than i2t's, when I use a bigger K, like...

I just use t2i re-rank directly without text-text fusion network, but the result improved so much. that is my result on t2i: original: 52.94 79.7 86.7 1.0 9.9886 t2i re-rank:...

thank you