Using the gait 2392 model for kinematic and kinetic analyses

[Bernhard Faerber]

Hi everyone,

I recently encountered a new issue when trying to locally reprocess a number of trials locally in order to use a different MS model.

When executing the reprocessSessions.py script I get the following output in my console:

Processing Dynamic_09
The pose pickles for OpenPose 1x736 already exist in the database. We will download them to avoid re-running pose estimation
Load extrinsics for Cam0 - already existing
Load extrinsics for Cam1 - already existing
Load extrinsics for Cam2 - already existing
Load extrinsics for Cam3 - already existing
Running OpenPose for Cam0
Running OpenPose for Cam1
Running OpenPose for Cam2
Running OpenPose for Cam3
Synchronizing Keypoints
Using general sync function.
You tried to filter 240 Hz signal with cutoff freq of 500. Will filter at 120 instead.
You tried to filter 240 Hz signal with cutoff freq of 500. Will filter at 120 instead.
You tried to filter 240 Hz signal with cutoff freq of 500. Will filter at 120 instead.
You tried to filter 240 Hz signal with cutoff freq of 500. Will filter at 120 instead.
For Cam1, used reprojection error minimization to sync.
For Cam2, used reprojection error minimization to sync.
For Cam3, used reprojection error minimization to sync.
Writing synchronized videos
Augmenting marker set
Using augmenter model: v0.3
1/1 [==============================] - 2s 2s/step
1/1 [==============================] - 2s 2s/step
tuple index out of range

I tried running the script several times, however the outcome does not change.
Currently all captured trials appear with red dots next to them in the web app (Session ID: 28a8110b-8399-41dd-9e9e-b944a3dbbc25).

How do I best solve this issue and can we expect to see a model with unlocked frontal plane knee ROM being implemented like with the ISB shoulder model for kinematic and kinetic analyses?

Thank you for your continued support.

Best

Bernhard

[Scott Uhlrich]

I recommend dropping some breakpoints in main.py. It is failing somewhere between https://github.com/stanfordnmbl/opencap ... in.py#L389 and https://github.com/stanfordnmbl/opencap ... in.py#L488

A model with knee adduction is on the TODO list, but I can't say when it will get done.

[Bernhard Faerber]

Thanks a lot Scott,

after setting a couple of breakpoints the issue seems to be ~ at line 480 of main.py, where the IK part of the OpenSim Pipeline occurs.

How do I best proceed from here?

Best

Bernhard

[Scott Uhlrich]

This may have to do with the different model that you're using. Not sure we can help debug much further than this without more specific error messages. Try isolating the line of code and getting the exact error message (drop a breakpoint right before the line, then hit f9 to run the line that is causing problems)

[Bernhard Faerber]

Hi everyone,

I managed to solve the problem by running the operation on a more powerful RTX 3080 setup.

Best,

Bernhard

[Antoine Falisse]

Great, thanks for letting us know.