I wonder if this is the right place for asking, but didn't know else where to...
I'm currently using Vimeo-90k triplet dataset, which specifies train-list / test-list images in tri-trainlist.txt and tri_testlist.txt each.
Though the number of train dataset to be used is about 50K, test dataset about being 4K. The rest(about 20K) are not used for training nor testing according to the text files given.
All the previous frame interpolation works(IFR-Net, FLAVR...) seem to have used tri_trainlist.txt for there training, so wondering, what about the rest of the dataset.
Wish if anyone can give me an explanation about this issue, I'd really appreciate it. If this is not the right place to ask this question, please let me know.