search

Moohnai / MOFO

The main contribution is to make self-supervised video representation learning more meaningful by raising awareness of motion data
7 stars 1 forks source link

When to open source code? #1

Closed leexinhao closed 1 year ago

Moohnai commented 1 year ago

Hi @leexinhao, the code is now available

leexinhao commented 1 year ago

Hi @leexinhao, the code is now available

Thanks for your nice work! And I would like to know whether you have conducted experiments on kinetics dataset or ViT-L. From several concurrent works using motion, it seems that the results on vit-L are not good.

Moohnai commented 1 year ago

Thanks @leexinhao. I only tried on SSV2 and Epic-Kitchens using ViT-Base.