search

jiaw-z / FCStereo

[CVPR'22] Revisiting Domain Generalized Stereo Matching Networks from a Feature Consistency Perspective
MIT License
53 stars 3 forks source link

About the generalize ability of GwcNet. #4

Closed gangweiX closed 6 months ago

gangweiX commented 2 years ago

I test generalize ability of GwcNet on KITTI and Middlebury using the official pretrained model, the result is 12.5% on KITTI 2012, 12.4% on KITTI 2015. But in your paper, it is 20.2% and 22.7%. Can I know how you tested?

jiaw-z commented 1 year ago

I test generalize ability of GwcNet on KITTI and Middlebury using the official pretrained model, the result is 12.5% on KITTI 2012, 12.4% on KITTI 2015. But in your paper, it is 20.2% and 22.7%. Can I know how you tested?

Very sorry to reply late. It puzzles us as well. It seems our retrained GWCNet gets a different generalization performance than the official pre-trained checkpoint despite we follow the original implementation. Since we observe similar results in other works(DSMNet...), we report the results in our experiments.

colorfulgreen commented 6 months ago

[NOTICE] Zhang Jiawei(Gary) deleted my workspace without notice, which containes data and network models of a paper under review.

image