Can we use group convolution over EDVR?

[leiwen83]

Hi,

EDVR is a amazing model, but its compution effort is large. Could we use some tech like group convolution to reduce the FLOPS? As https://github.com/DestinyK/MobileSR shows it use "parallel group convolution" to replace original convolution layer.

Not sure whether someone has tried this before?

Thx, Lei