search

feihuzhang / DSMNet

Domain-invariant Stereo Matching Networks
MIT License
227 stars 28 forks source link

Settings for SGM #7

Open mli0603 opened 3 years ago

mli0603 commented 3 years ago

Hello:

Thanks so much for open-sourcing the code. The presentation also looks nice!

I wonder what settings did you use for your reported result for SGM, because there are lots of parameters one can tune. Is this an already implemented version from libraries such as OpenCV, or is it a customized version? Do you mind sharing more details? Thank you!

ZhiboRao commented 3 years ago

Hello，I'm not an author in DSM-Net, but we find a project about SGM based on openCV (python or cuda version) that can work better than results in DSM-Net.
you can find it in https://github.com/beaupreda/semi-global-matching

However, it is very slow (50min for one image), I suggest you use the CUDA version. (100fps for kitti in 1080TI) you can find it in https://github.com/fixstars/libSGM

if you need the CUDA version with a python interface that we add a python interface in libSGM (70fps for kitti in 1080TI). you can find it in our organization https://github.com/Archaic-Atom/libSGM

I hope it can help you.