code & paper inconsistant

[ymclsh]

Hi

Thanks for sharing the code.
I was confused at several places in this code as inconsistent with the paper, until I read another paper QNS.
I think this code is kind hybrid-implementation for these two papers, right?
But why did you that? Would it be more clear to test upon pure implementation for ONE paper to verify the advantages and disadvantages of the methods proposed in that paper?

Regards, ymcl

[yaoyx689]

Hi,
As mentioned in the paper, this method is an extension of our CVPR paper, hence some components are similar. Could you please point out which part of the code does not match the paper? We will check it out.