Catalogs of minimal problems from mathematicians for computer vision engineers — How to make discrete data FAIR across disciplines?

This talk will provide more questions than answers. We will discuss FAIR discrete data and how we should / can deal with such data in our everyday research. How should we store discrete data if there is no agreed-upon standard format? Where should the data be stored? Who should maintain the database? How can the data be found and used by researchers outside our area of expertise?

I will discuss these questions following a self-critical example: Together with fellow mathematicians, we compiled two lists of point-line arrangements (with 140616 resp 74575 entries) that encode efficiently-solvable 3D-reconstruction problems. Those so-called ‘minimal problems’ are a key ingredient for computer vision engineers who develop 3D-reconstruction algorithms. It would be our dream if our lists of minimal problems would be used in practice by engineers, and for that an early reflection about the FAIRness of our data would have helped.

The computer-vision part of this talk is based on joint work with Timothy Duff, Anton Leykin, and Tomas Pajdla.