[SUGGESTION] Zooming-Slow-Mo implementation

victorca25 / traiNNer

traiNNer: Deep learning framework for image and video super-resolution, restoration and image-to-image translation, for training and testing.

Apache License 2.0

287 stars 40 forks source link

[SUGGESTION] Zooming-Slow-Mo implementation #33

Open MarcoRavich opened 3 years ago

MarcoRavich commented 3 years ago

Very cool fork...

According to this video they achieved better (quality) performances of DAIN+EDVR, so it would be great to be implemented in standard-motion too:

https://github.com/Mukosame/Zooming-Slow-Mo-CVPR-2020

Hope that inspires !

victorca25 commented 3 years ago

Thanks for the suggestion! One limitation for integrating ZSM was, similar to EDVR, the dependence on custom CUDA kernels for the Deformable Convolutions v2, but since it looks like it will be soon integrated in a new version of mainline PyTorch, it could be added as an option soon!