RMSE Output when using No Fit Physics?

Upload mocap files and get an optimally scaled OpenSim model and IK back in minutes. Share your data with the community. Browse and download biomechanics data. Hosted by Stanford University.

Available at https://addbiomechanics.org
POST REPLY
User avatar
Chloe Baratta
Posts: 3
Joined: Fri Aug 28, 2020 6:20 am

RMSE Output when using No Fit Physics?

Post by Chloe Baratta » Tue Feb 13, 2024 7:32 am

Hi,
I noticed that the processing log only outputs average marker RMSE when Fit Physics is enabled. Is there any way to a) also report max marker error and b) report average marker RMSE/max marker error when not using Fit Physics? I remember this being built into the GUI in summer 2022, so not sure if I'm just not looking in the right place.
Thanks!
Chloe

User avatar
Nicholas Bianco
Posts: 1027
Joined: Thu Oct 04, 2012 8:09 pm

Re: RMSE Output when using No Fit Physics?

Post by Nicholas Bianco » Tue Feb 13, 2024 4:30 pm

Hi Chloe,

I just tested this on sample data and Marker RMSE still appears when skipping the physics fit. Note that the results are now printed at the bottom of the subject page.

If you provide a link to your subject page I can take a look (if you're still encountering an error).

Best,
Nick

User avatar
Chloe Baratta
Posts: 3
Joined: Fri Aug 28, 2020 6:20 am

Re: RMSE Output when using No Fit Physics?

Post by Chloe Baratta » Thu Feb 15, 2024 12:54 pm

Hi Nick,
Yeah - I can see the marker RMSE for each trial - no issues there thankfully. If I download the processing log for a model with Fit Physics enabled, there's a line with "Avg Marker RMSE: X.X cm". Is this not the scaling RMSE? To rephrase, I guess I'm asking if there's any way to get scaling error reported, and then also max marker error during IK to output.
Thanks,
Chloe

User avatar
Nicholas Bianco
Posts: 1027
Joined: Thu Oct 04, 2012 8:09 pm

Re: RMSE Output when using No Fit Physics?

Post by Nicholas Bianco » Thu Feb 15, 2024 4:36 pm

Hi Chloe,

Apologies, I see that you're referring to the log file (not the info logged to GUI). If that line is only printed during dynamics fitting then there's probably some inconsistencies recently introduced in the logging behavior. We'll take a look into it.

For now, you should be able to use the "B3D Summary Statistics" at the bottom of the log file to get average marker RMSE (it's labeled "Marker RMS" there). For example:

Code: Select all

B3D Summary Statistics:
  Num Trials: 1
  Num Processing Passes: 2
  Num Dofs: 37
  Trial 0:
    Name: walk_segment_0
    Num have GRF frames: 126
    Num missing GRF frames: 0
    Processing pass 0:
      Marker RMS: 0.016066611315759403
      Marker Max: 0.045487796207742084
      Linear Residual (on frames with GRF): 1192.66504890688
      Angular Residual (on frames with GRF): 109.11076510337091
    Processing pass 1:
      Marker RMS: 0.016067086245923762
      Marker Max: 0.04548899515990227
      Linear Residual (on frames with GRF): 1221.3822454637098
      Angular Residual (on frames with GRF): 90.51276660734608
Best,
Nick

POST REPLY