Trainee Support 2023 08 23

This is the first “help desk” meeting to get issues from the trainees. Items raised here may end up in the B4K backlog to be worked on by staff or anyone else that has time to help out.

Attendees: Eli Kinney-Lang, Greg Wilding, Daniel Comaduran Marquez, Vella Kim, Ludy Borges, Araz Minhas, Masuma Akter, Amy Weiler, Anup Tuladhar

Discussion

Eli Kinney-Lang described the purpose of the meeting, answered questions.

Does this have to do with only software issues? Nope, whatever trainees need is fair game

Should you prepare for this meeting with items? Yes, that makes the meeting work.

Python scripting tips

Vella asked about how to organize the various Python scripts that will be needed for her project. Discussion ensued about how to organize the Python code, files and how to structure functions. The team talked about what’s worked for their projects.

TODO - May want to provide help with modular style programming. Simple organization, ex: breaking large functions into helpers is a nice way to promote reuse within your own project.

Rejecting bad data / results

Vella asked about identifying bad results and rejecting them. There some discussion about the right approaches (automated, procedural, manually).

Eli pointed out that manual preprocessing because it’s hard to repeat - main concern is rejecting things until you get the results you want. Automated tools (filtering) aren’t perfect, but will be more consistent (Dan has a tool, ICA, Riemannian potato fields, etc). Recommendation is to apply automation, then look at results and see what you get.

EEG Visualization

Vela asked about “the script that popup up offline visualizations for data”. There is a spike for this in the backlog. Jupyter notebooks is an option (modern version of gnuplot), though these aren’t interactive.

Eli - the biggest problem they have is getting the data into the visualization tool, not making the picture. This is always going to be a somewhat custom job for each project, though we could provide a tutorial on a common way to approach.

Importing media pipe

Ludy raised an issue she’s having with importing mediapipe (?). Needed to use an older version of Python to make it work (Grace suggested this). [not sure I got this right]

There was a discussion about face landmarks. Eli has an example to share that just works (said it was a legacy solution). Discussion ensued. Should be easier than what Ludy was experiencing.

EEG data set

Masuma needs one set of EEG / motor imagery data to start using the pipeline [?]. Doesn’t sound like it has to be anything specific, just something to see if the tool work. Eli has some data in XDF format.

Autodesk Fusion Woe

Dan shared his issue with some mechanical models for Baccia (Autodesk Fusion 360). Wanted input on how to organize the flat mess of files that he’s got in the editor. Sometimes it helps to share the problem out loud and see what the group says, several ideas were shared.