search

jamycheung / DELIVER

Repository of DELIVER dataset and CMNeXt models (CVPR 2023)
Apache License 2.0
129 stars 6 forks source link

Evaluation issue #17

Open 800166 opened 2 months ago

800166 commented 2 months ago

We are impressed by your work. The proposed method achieved 66.3 mIoU on DeLiVER. However, I am confused that 66.3 is achieved on validation or testing subset? I am looking forward to your reply. Thank you!

iacopo97 commented 1 month ago

I think it is acheived on the validation set. I have tried their val_mm.py and I scored 66.3 on the validation set

800166 commented 1 month ago

I think it is acheived on the validation set. I have tried their val_mm.py and I scored 66.3 on the validation set

Thank you so much! But I'm also a bit curious why the authors did not list the results on the test set, which should normally be given in the paper.