Hilti-Research/hilti-slam-challenge-2021

Submission requirements clarification

Opened this issue · 4 comments

Hi Hilti team,

Just some clarification on this line from the readme: "Each participant should submit the estimated trajectories for the above datasets and a report describing the adopted method."

Our SLAM produces best results for captures with an initial static calibration period. As other contributors have seen, not all of the datasets have been recorded in this way.

Are we required to submit trajectories for all datasets uploaded, or are we free to focus on the most suitable captures?

Many thanks

you can focus on the most suitable one of course, however for the ranking the error on all trajectories is going to be evaluated (the final score will be a combination of rmse error and completeness of the trajectories).

@hemi86 Thanks for your response.

"the ranking the error on all trajectories is going to be evaluated" - do you mean all trajectories submitted by us, all trajectories derived (by yourselves) from the datasets you have uploaded?

sorry for the confusion, i wanted to say the the error on all datasets is going to be evaluated. If you do not submit a trajectory or only a part of a trajectory, it will lower your score on completeness.

sorry for the confusion, i wanted to say the the error on all datasets is going to be evaluated. If you do not submit a trajectory or only a part of a trajectory, it will lower your score on completeness.

Hi hemi86,

Does the estimation rate affect completeness? For example if I only publish estimate at half the camera rate, would it affect the score compared to publishing it at full rate?