[ENH] Add various online datasources to the catalog (plugins)

[edwardchalstrey1]

Feature Request

Make various online data sources easily loadable via scivision

Describe the solution you'd like

Similar to #218 but for datasources instead of models

see: workshop notes

• [ ] Write a "how to build a scivision plugin" tutorial for a new online datasource

[quantumjot]

These would be a good to add:

• [ ] https://idr.openmicroscopy.org/cell/
• [ ] https://www.rxrx.ai/rxrx1#the-data

[edwardchalstrey1]

~~Also things like mnist - default image datasets~~ boring, let's stick to scientific imagery

[ots22]

• Miquel's csv-image intake driver would allow a few to be added easily (e.g. from squidle)
• Selected datasets from the Polar Data Centre (discussion with Paul B)

[acocac]

Earth Observation

• Radiant MLHub

[edwardchalstrey1]

@quantumjot for the IDR data for example, what would you envision being in the scivision catalog? Do you think a catalog data entry that people can find via a query with a model trained on cell or tissue data, but which only has data for a specific example experiment would suffice? That way the user gets to discover the resource, and has some test data to play with, then they can go on to do things outside of scivision?

[edwardchalstrey1]

Idea from todays group session:

• Add catalogs other than the default that come from online e.g. https://stacspec.org/en/tutorials/access-sentinel-2-data-aws/ which can be opened with: https://github.com/opendatacube/odc-stac
• Would be great if there were similar examples from other domains

[edwardchalstrey1]

https://iiif.io/

[edwardchalstrey1]

another example: https://www.quantitative-plant.org/dataset

[ots22]

https://openorganelle.janelia.org/