You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi. I was trying to reproduce the performance of PSPnet on cityscapes using the catalyst training framework, which aims to facilitate reproducible and easy-to-do deep learning research.
After some try, I can only obtain mIoU of 0.739, which is a big gap from the 0.773 provided by this repo.
I was wondering if any of the authors is willing to help me to obtain the optimal performance of PSPNet on cityscapes on catalyst framework?
The training code based on this repo and catalyst framework can be found here
Thanks in advance.
The text was updated successfully, but these errors were encountered:
Hi. I was trying to reproduce the performance of PSPnet on cityscapes using the catalyst training framework, which aims to facilitate reproducible and easy-to-do deep learning research.
After some try, I can only obtain mIoU of 0.739, which is a big gap from the 0.773 provided by this repo.
I was wondering if any of the authors is willing to help me to obtain the optimal performance of PSPNet on cityscapes on catalyst framework?
The training code based on this repo and catalyst framework can be found here
Thanks in advance.
The text was updated successfully, but these errors were encountered: