First RFC Lab Residence - ML on soil chromatographies (Looking for collaborators)

We are very happy to announce that the first Residence at the RFC Lab is happening on mid July. @leandro.mastrantonio is going to visit @gbathree to run around 400 soil chromatographies from the RFCLab soil library. These samples (with lots of metadata) have already been tested for several parameters like organic carbon, mineral content, and surface reflectance with the Our-Sci reflectometer.

The idea is to use this dataset (chromas+standard test) to work on a quantitative proof of concept. We will apply image analysis tools to read/extract the colors, patterns and features of the chromas and then use ML algorithms to correlate this results with the standard tests data. For this part we would love to have the advice/help of some AI experienced person (@spmohanty @vektor @Ipppolit @kaspar @bengtsjolen just to pin a few of you out there).

This is a follow up from the chroma workshop made at Shenzhen

There are a few previous considerations about the ML point in this previous thread.

More info and experimental plan of the residence here.

The fresh outcome of the residence will arrive just in time for GL GOSH so I think discussion will continue there.

Please feel free to pin anyone who you think might be interested on this


Hey @nanocastro!

This sounds like a great project. Unfortunately Iā€™m not very AI experienced. I just played around with some computer vision libraries so far but if you need help e.g. to manually annotate the images for an AI/ML training dataset or other tasks I can help.


Hi Alex
Thanks for the interest and enthusiasm. It would be great if you can collaborate with us. Maybe we can arrange a little chat so we can better understand how.
I will pm you this week