Post

@cppape.bsky.social

I believe that more diverse and high quality data is the most important thing we need for creating better AI models for bioimage analysis. That's why we proposed BioImageNet, and now got it funded!

NFDI4BIOIMAGE@nfdi4bioimage.bsky.social

We are glad to announce that the @dfg.de has funded our #BioImageNet proposal! 🤩

Adrian Wanner

@adrianawanner.bsky.social

Will there be dedicated funding to acquire these datasets?

Constantin Pape

@cppape.bsky.social

No, unfortunately that is out-of-scope for the funding call. It's dedicated funding to create data corpora based on existing data. Our main goals are: i) (meta-)data standardization (based on OME-Zarr work), ii) quantifying how a given dataset affects (foundation) model training,

22/10/26, 8:27 am0 reposts1 like

Constantin Pape

@cppape.bsky.social

e.g. for segmentation models or VLMs, iii) building a FAIR data corpus that is extensible, including data provenance and credit for data providers.