High Fidelity Solver

[Krish Narotam]

Hi,

I am having trouble with our patients data. they ended up very choppy and they constantly glitch and I am pretty sure the set up is all optimal.

This is our session: 52def0cd-e3e7-4a60-90d0-1981d9159e89

I remember hearing about a "High fidelity machine learning" solver or something that could help reprocess the video data to potentially smooth out and make the data more accurate.

I know that OpenCap hasn't been validated for prosthetics (which is something we are investigating). But I figured a different solver might fix some of the issues that we are having.

Thanks.

-Krish

[Krish Narotam]

if you look in "walk2" you can see part of my issue.

I forgot to put my question in, but if there is way to run it with a slower but maybe more robust solver, if you could point in me in the right direction, that would be helpful.

-Krish

[Matt Petrucci]

Hi Krish, I think you may be referring to the newer augmenter v0.3, which you used, but there are two things you can try that might clean up your trials. They are in the best practices here: https://www.opencap.ai/best-practices

1. Try the hand punch at any point in the trial to help with synchronization.
2. Also try and allow for more time before the person starts moving (about 1 second). In walk2 the person starts walking right at the start of the recording.