charlesCXK / TorchSemiSeg

[CVPR 2021] CPS: Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision
MIT License
504 stars 74 forks source link

Question about the crop size #23

Open revaeb opened 3 years ago

revaeb commented 3 years ago

Thx for your marvelous work! When I was trying to reproduce your work in voc2012 dataset, I felt strange about the size of inputting images after cropping operation, should it be like batch channel321321, same with other methods? In your work the size of images is batch channel512512 now,if I have not made a mistake.

charlesCXK commented 3 years ago

Hi, we use the crop size 512x512, as illustrated in the paper. We reproduce other methods under the same training setting including the crop size (an example here), so the comparisons are fair. Some papers use the crop size 321x321 but we haven't tried that. I think the experimental phenomenon will be the same (or similar) using different crop sizes.

revaeb commented 3 years ago

Appreciate for your quick response! I'll check it out!

JihwanEom commented 3 years ago

@revaeb Hello, did you run experiments with 321x321 crop size? Can I get some results about this?