search

Parskatt / RoMa

[CVPR 2024] RoMa: Robust Dense Feature Matching; RoMa is the robust dense feature matcher capable of estimating pixel-dense warps and reliable certainties for almost any image pair.
https://parskatt.github.io/RoMa/
MIT License
622 stars 51 forks source link

Coarse vs upsample res #70

Open nonopx opened 3 months ago

nonopx commented 3 months ago

Hey, I saw the default values for the coarse is 560×560 and for the upsample is 864×864

I'm using 1000×1000 images, do you think it is okay to keep the default values or change them ? For some of my images the models find arbitrary correspondenses. Thank you very much

Parskatt commented 3 months ago

I suggest you visualize the warp and confidence before the sampling, try some different resolutions and see what works best.