Bioimaging knowledge share

[metazool]

We've been invited to participate in a bioimaging knowledge share workshop, apparently this is a spin-out from a recent Turing AI-UK special interest session. This includes an opportunity to offer a 10 minute presentation

@albags has offered a longer talk recently covering the work on the AMI trap (moth light plus camera, v2 with a bioacoustic sensor). I'm happy to undertake to edit that down a little; I'd like to add a couple of slides about other recent use cases and approaches, e.g. the vector embeddings search of plankton images and with permission from the researcher, the segmentation approach in #12

[metazool]

_Here are my notes, they got briefer as the discussion became more interesting, glad we went and thank you again for the presentation @albags _

Knowledge graph based reasoning and ontology, explainability

Not necessarily thinking about physical sampling complementary to cameras

Sophie Pitois @ Cefas to follow up with

"richer contextual feedback for citizen scientists"

Oliver Strickson "what next", Tom A - futures conversation within UKCEH

Overview of the process inside the flow cytometer and its raw output
Raw bands / histograms / spectrograms fom the flow cytometer - it's not a vision problem, you could use a spectral diffusion model for it as well

.cyz files for the output, unpredictable and undocumented processing of the proprietary libs

CytoClus cyz2json, @LucindaLanoy

https://github.com/OBAMANEXT/cyz2json

OBAMANEXT/cyz2json#4

"cytometric signals"

"indicator species"

(discussion about classification problems in general)

What is it uniquely that this group can do says Tom A

"morphologically based traits" without doing counting or classification

The flow cytometer is already, intrinsically taking multimodal observations (spectral-ish and imaging)