search

Mukosame / Zooming-Slow-Mo-CVPR-2020

Fast and Accurate One-Stage Space-Time Video Super-Resolution (accepted in CVPR 2020)
GNU General Public License v3.0
915 stars 164 forks source link

Number of deformable groups used in Sakuya_arch.py #26

Closed sourabh-patil closed 4 years ago

sourabh-patil commented 4 years ago

In Sakuya_arch.py file, to execute deformable convolutions, authors are using numbers of groups as 8. So is it number of groups that we use in normal convolutions? If it is so then is there any reason to keep it as 8? Is it the same concept that was used in AlexNet paper to increase the computational efficiency or there is some other usage too?

YapengTian commented 4 years ago

It is not for computational efficiency. The trick is firstly used in TDAN. With more groups of deformable Conv, the model will have a higher capacity to explore temporal contexts with sampling more different positions. It will improve temporal alignment performance.

sourabh-patil commented 4 years ago

@YapengTian Thank you for the swift reply and thoughtful explanation!