How to generate object-like masks
sczhou opened this issue · 2 comments
sczhou commented
Hi authors,
Thanks for your awesome work!
I'm wondering if you used the same 'create_random_shape_with_random_motion' function for both video completion and object removal, if so, can I say this model has only been trained once for both tasks?
Besides, does this moving mask (https://github.com/MCG-NKU/E2FGVI/blob/master/core/utils.py#L209) refer to the object-like masks mentioned in your paper (experiment settings)?
Paper99 commented
Hello, your understanding is correct. We only trained this model once for both tasks. Training masks are generated by the 'create_random_shape_with_random_motion' function.
sczhou commented
Well noted. Thank you!