sea-shunned

I've been supporting non-computational (i.e. scientists) to use and finetune SAM for biological applications, so excited to see how SAM2 performs and how the video aspects work for large image stacks of 3D objects.

Considering the instant flood of noisy issues/PRs on the repo and the limited fix/update support on SAM, are there plans/buy-in for support of SAM2 on the medium-term beyond quick fixes? Either way, thank you to the team for your work on this and the continued public releases!

I agree that adding state pension would be useful, though that's another layer of maintenance...

Real slick though and very clear, thanks for putting it together! Would you mind sharing a bit about what you're using underneath, as this doesn't look like the usual Shiny-type interface.

Streamlit[0] was created specifically to create dashboards for ML/data science groups, and I've found it pretty useful. I've used it for research (model inspection and development), as well as teaching and it's been pretty useful for that.

[0] - https://streamlit.io/

I'll agree that there is increasing emphasis on reproducibility and _useable_ software in academia. Writing documentation, unit tests etc. is still not really rewarded properly, but at least within the current paradigm such efforts are often rewarded with more users (and therefore citations) which is rewarded. Soon, hopefully, it'll be recognised more directly.

Also, I'm currently a postdoc in medical imaging at UCL, super interested in learning a little more about the group in Oxford you mentioned if you're OK with sharing a link/group name? I may be able to guess but just want to check!

It's also a shame that this paper in turn doesn't cite "Testing Heuristics: We Have It All Wrong" (J. N. Hooker, 1995) (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.71....), which discusses these issues in much the same way. It's good to see in "The Benchmark Lottery", however, that they look more into specific tasks and their algorithmic rankings and provide some sound recommendations.

One thing that I'd add (somewhat selfishly as it relates to my PhD work), is the idea of generating datasets that are deliberately challenging for different algorithms. Scale this across a test suite of algorithms, and their relative strengths and weaknesses become clearer. The caveat here is that it requires having a set of measures that quantify different types of problem difficulty, which depending on the task/domain can range from well-defined to near-impossible.