RoMa/README.md at main · Parskatt/RoMa · GitHub
RoMa is a robust dense feature matcher for estimating pixel-dense warps and certainties in image pairs, with demos available.
Issues · Parskatt/RoMa · GitHub
How to adjust the coarse_res and upsample_res parameters to the optimal state? #110
How to adjust the coarse_res and upsample_res parameters to
I recently found that slightly modifying the coarse_res and upsample_res parameters will cause a sharp decrease in the number of matching points when running RoMa for matching.
Matches and Config: · Issue #74 · Parskatt/RoMa
Hi, thanks for this great repo. I was wondring from where i can change max number of features and matches that model will detect and try to match, similarly where is the threshold value
GradScaler growth_interval parameter · Issue #109
Why do you set the GradScaler growth_interval=1e6? Is this just randomly chosen, or where you facing some stability issues with the default parameter
Value of ''coarse_res'' and ''upsample_res'' for ScanNet
Would it be possible to get the parameters used to generate the results? I am getting quite poor results for keypoint matching on the ScanNet test case
Training set and test set setting? · Issue #54 · Parskatt/RoMa
Hi Author, Do you use the megadepth dataset from 0000 to 5014 as training set like the 1st and 2nd images show? 2.When prep_info''s corresponding Undistorted_sfm has both dense0 and
How does T_1to2 generated? · Issue #48 · Parskatt/RoMa
I am doing some epipolar work, however, I found that rotational part of T_1to2 differs much from what I solved from matching pairs. And sampson distance calculated from T_1to2 and
RoMa/README.md at main · Parskatt/RoMa · GitHub
[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. -
export onnx file issue · Issue #130 · Parskatt/RoMa
Notifications You must be signed in to change notification settings Fork 123