Hi!
First of all, thanks for your great work! It's very important and challenging to evaluate video inpainting models. It's greatly appreciated that your work has created the an important benchmarking dataset and platform to do that.
One issue encountered is that: when downloading any of the split or the whole Flickr video dataset, the [links] (https://web.eecs.umich.edu/~szetor/media/DEVIL/datasets/devil/flickr-all.tar) seems to be broken. Could you please take a look at it?
Thanks!