arclab-hku / ESVIO

ESVIO: Event-based Stereo Visual Inertial Odometry
https://github.com/arclab-hku/Event_based_VO-VIO-SLAM
GNU General Public License v3.0
25 stars 3 forks source link

The evaluate method/command #5

Closed DayBeha closed 1 day ago

DayBeha commented 2 weeks ago

Hey, thank you for your excellent work! But I wonder what command you use for evaluation?

I downloaded the results in https://github.com/arclab-hku/Event_based_VO-VIO-SLAM/blob/main/Results_for_comparison.md and tried like evo_ape bag ESVIO_VECtor_desk_normal.bag /gt/pose /pose_graph/evio_odometry -va -p, but can not get the value of MPE and MRE described in your paper.

The output is:

Opening bag file ESVIO_VECtor_desk_normal.bag
Loaded 10321 geometry_msgs/msg/PoseStamped messages of topic: /gt/pose
Loaded 17205 nav_msgs/msg/Odometry messages of topic: /pose_graph/evio_odometry
--------------------------------------------------------------------------------
Synchronizing trajectories...
Found 10321 of max. 10321 possible matching timestamps between...
    /gt/pose
and:    /pose_graph/evio_odometry
..with max. time diff.: 0.01 (s) and time offset: 0.0 (s).
--------------------------------------------------------------------------------
Aligning using Umeyama's method...
Rotation of alignment:
[[-6.93314624e-01  7.20205748e-01 -2.48699329e-02]
 [-3.66840719e-02 -8.05906295e-04  9.99326588e-01]
 [ 7.19700710e-01  6.93760068e-01  2.69788261e-02]]
Translation of alignment:
[-0.74994847  1.3441179   0.27788658]
Scale correction: 1.0
--------------------------------------------------------------------------------
Compared 10321 absolute pose pairs.
Calculating APE for translation part pose relation...
--------------------------------------------------------------------------------
APE w.r.t. translation part (m)
(with SE(3) Umeyama alignment)

       max  0.121161
      mean  0.051990
    median  0.048060
       min  0.004991
      rmse  0.060048
       sse  37.214805
       std  0.030047
cpymaple commented 2 weeks ago

Hi, for the calculation of MPE/MRE, you can refer to the following issue: https://github.com/arclab-hku/Event_based_VO-VIO-SLAM/issues/5