Johan Edstedt
Johan Edstedt
We have wheels on pypi (pip install poselib), is there a usecase for also having a conda dist?
This also shows in the variance of different ransac runs. With regular gcransac Im getting low variance between runs, while the affine estimates affine ransac gives me much higher variance.
See https://github.com/pytorch/pytorch/issues/121834
Hi, that's great news! My framework is mostly just educational (and trying to keep it as not too many lines), so I don't expect it to outperform proper bundle, but...
Hi @royw99 , sorry for the late response. In general, COLMAP relies on keypoint correspondences. By default RoMa randomly samples a set of correspondences between each pair, which will not...
Have you had a look at the [demos](https://github.com/Parskatt/RoMa/tree/main/demo)
Hi, I think I used the default values as in roma_indoor. Although it could be the case that I didn't use upsample_preds, I don't quite remember.
If you run the model (without keypoints), are you able to get results comparable to our reported results?
No demo yet, but see here: https://github.com/Parskatt/RoMa/blob/50522299a55efc14ed892caaca9e29a1c8b73e12/roma/models/matcher.py#L545 Note: assumes keypoint coords in normalized [-1,1] grid, and we use colmap conventions for pixel coordinates (i.e. you might want to do +0.5...
> Thanks, could you please share what the expected shapes of `x_A` and `x_B`? Are they expected to be the number of samples x 2? Yes, and it seems I...