Question on the bilateral filter approximation

Hi,

Thanks for sharing the code!

By reading the code I was unable to see which approximation you use for the `BilateralFilter`. It doesn't seem to be the permutohedral lattice as used in the CRFasRNN paper. Could you please share a bit more detail on this?

Thank you in advance for your reply.

Best regards.

