Why use 5 different resolutions in the same convolutional refiner

[skill-diver]

Hi, Author,

Thank you for the sharing. I am confused about why you use 5 different resolutions and the same convolutional network. And why you choose this convolutional network architecture?

[Parskatt]

Hi, could you be a bit more precise? The refiners use a coarse to fine approach which is common in matching tasks.

[skill-diver]

Yes, what I mean is why you choose 16 8 4 2 1 as the scale repeatedly in a same conv_refiner, why not just choose a network which could accept all resolutions input?

[Parskatt]

Typically you get worse performance that way, you can use more channels at lower resolution. If you use a single network that's difficult.

[skill-diver]

Thank you. So, you need a different input channels setting for different resolutions. Do you think is there a powerful network which could use just one parameter setting to do the good work like the different scales settings mulitple convolutional refiner now?

[Parskatt]

Not impossible, but I'm not sure what the benefit would be.