evaluation

[knelk]

Hi, thanks for the amazing work.

Can you please clarify how do you compute your error metrics, exactly?

Tab. V shows the relative error. Do you use the RPE-script from https://cvg.cit.tum.de/data/datasets/rgbd-dataset/tools#evaluation? If yes, please specify how you call the script (which parameters), but mainly how do you set --max_pairs?

Also, do you do any post-/preprocessing to the trajectory, e.g. interpolation,...?