GPS vs 3D error in quality report

I’ve searched around the opensfm docs as well as here and haven’t found a clear answer on how one would describe these two error categories. Could anyone share a bit of context of GPS vs x/y/z. Is GPS the drone position error and 3D the error in the point cloud?


@Saijin_Naib, any ideas here?

GPS error is a measure of how much you can expect the georeferencing accuracy to be, relative to the GPS values of the photos (strange you have a value of 0 for the mean, what settings did you use?).

3D error is a measure of the relative accuracy of the reconstruction, in other words, if you reconstructed a measuring tape 1 meter long, and your 3D RMS is 0.044, you can expect the measuring tape’s point cloud and 3D model to be 1 ± 0.044 meters.long.

1 Like

Thank you @pierotofy! That makes sense, thank you for the response. I did think the mean of 0 was strange too which stimulated my question. I believe in this dataset I used gps-accuracy 0.1 (10cm). Flying the M300 RTK drone referenced to a known base point.

Actually I went back and looked. In this particular dataset I removed the gps-accuracy option…

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.