about the uploaded trained model

[siyueyu]

Hi, Jing,
Thank you for sharing the wonderful work.
I have tested the trained model you uploaded. But the results are different from the paper. Is the uploaded model your final result of your paper?

[JingZhang617]

Hi, This is the final model, which leads to performance slightly different from what we report as no densecrf is used (one round of training). You can achieve better performance with an extra densecrf as refinement, and treat it as pseudo label for the second round training. Cheers, Jing

On Fri, 26 Jun 2020 at 1:00 pm, siyueyu ***@***.***> wrote: Hi, Jing, Thank you for sharing the wonderful work. I have tested the trained model you uploaded. But the results are different from the paper. Is the uploaded model your final result of your paper? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AE6B4FZOSGHFN6YIBIEHA7LRYQFMFANCNFSM4OI5ZWNA> .

-- Jing Zhang Ph.D. Student College of Engineering and Computer Science, Australian National University. Email: zjnwpu@gmail.com

[siyueyu]

@JingZhang617 Thank you for your quick response!