scaelles / DEXTR-PyTorch

Deep Extreme Cut http://www.vision.ee.ethz.ch/~cvlsegmentation/dextr

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Cannot reproduce the same result as provided Dextr_pascal-sbd model

Valentine233 opened this issue · comments

Thanks for your great work!

When I reproduce your results of Dextr_pascal-sbd, I cannot get the same performance (IoU=91.5).

Firstly, I just evaluate with your provided model "dextr_pascal-sbd.pth" and I can get the IoU of 91.5. However, when I try to train the model by myself, I only get the IoU of 86.2, with a gap of 5.3. I have barely changed your codes.

Here is my setup:

  • python 3.6.6
  • pytorch 0.4.0
  • gpu Titan X

Could you please give me some suggestions :)

Hi, could you try with different thresholds during evaluation? We have checked reproducibility at least 3 times, for different versions of PyTorch. Also, default behaviour of upsampling has changed (aligned corners), which should be fine if you stick to the given code.

Thanks for your suggestion.

Changing thresholds is helpful!

Hi, let us know whether this was the only problem.