APE and RPE error metrics

[majdros]

Thank you very much for the excellent work @MichaelGrupp . When evaluating APE and RPE, I noticed a significant deviation in the z-axis. However, I am only interested in evaluating the x- and y-axes, and I would like to restrict the evaluation to these axes without considering the z-axis. Is this possible?

[MichaelGrupp]

It's possible, but not from the command line without modifications. The fastest is probably to just set z to zero in your data. See also: https://github.com/MichaelGrupp/evo/issues/537

[mitchallain]

@MichaelGrupp are you open to a PR on this? Something like -r xy_trans_part to suppress any error along the z-axis?

[July0928]

i only see the rpe of translation , how can i get the results of Rotation ?

evo_rpe kitti KITTI_00_gt.txt KITTI_00_ORB.txt -va --plot --plot_mode xz --save_results results/ORB.zip

---

Loaded 4541 poses from: KITTI_00_gt.txt Loaded 4541 poses from: KITTI_00_ORB.txt

Aligning using Umeyama's method... Rotation of alignment: [[ 0.99983853 0.00400932 0.01751664] [-0.00361575 0.9997416 -0.02244238] [-0.01760209 0.02237542 0.99959467]] Translation of alignment: [-1.32278266 0.31999263 3.31982374] Scale correction: 1.0

Found 4540 pairs with delta 1 (frames) among 4541 poses using consecutive pairs. Compared 4540 relative pose pairs, delta = 1 (frames) with consecutive pairs. Calculating RPE for translation part pose relation...

RPE w.r.t. translation part (m) for delta = 1 (frames) using consecutive pairs (with SE(3) Umeyama alignment)

   max      0.302712
  mean      0.019301
median      0.014709
   min      0.000312
  rmse      0.028120
   sse      3.590030
   std      0.020450

[mitchallain]

I believe this was addressed by #607, which allows the pre-projection of trajectory data onto a plane.

[MichaelGrupp]

Yes, it's documented here: https://github.com/MichaelGrupp/evo/wiki/Metrics#projection