Danforth Center Awarded NSF Grant to Build Data-Sharing Ecosystem for Plant Image Research

New project connects three open-source tools to solve a longstanding data-sharing problem for plant scientists worldwide 

ST. LOUIS, MO (August 13, 2026) — The Donald Danforth Plant Science Center has received a $599,517 grant from the National Science Foundation (NSF) to build a data-sharing ecosystem for plant science image data, addressing a gap that has slowed research and machine learning development across the field. The grant comes as the NSF is on track to make its lowest number of awards in more than fifty years, according to the website Grant Witness. 

The three-year award, funded through NSF’s Findable, Accessible, Interoperable, Reusable Open Science (FAIROS) program and managed by the Office of Advanced Cyberinfrastructure, is co-led by Danforth Center Principal Investigator Malia Gehan, PhD, and Danforth Center Director of Data Science Noah Fahlgren, PhD. 

With this award, the Danforth Center aims to address a longstanding problem. When genomics emerged as a field decades ago, the federal government built public repositories, such as the Sequence Read Archive, so that researchers and companies across the world could reuse publicly funded DNA sequence data. No equivalent has existed for the images plant scientists collect, from microscopy of individual cells to whole-field drone flights, leaving valuable datasets scattered, hard to find, and difficult to reuse—a gap that has become more pressing as AI and machine learning models increasingly depend on large, well-labeled image datasets to train on. 

The project, called FAIR-FAIR (Findable, Accessible, Interoperable, Reproducible Facilitation of Accessible Imaging Research for Plant Phenotyping and Beyond), will connect three established open-source tools rather than build a new system from scratch: PlantCV, the plant image analysis toolkit developed at the Danforth Center; COPO, a metadata brokering platform built at the Earlham Institute in the United Kingdom; and the BioImage Archive, a public image repository operated by the European Bioinformatics Institute (EMBL-EBI). Together, the tools will let researchers analyze image data, describe it using community metadata standards, and deposit it into a public archive, all within a single workflow. 

“Right now, a researcher who wants to share plant image data publicly has to cobble together their own solution, and most repositories weren't built to handle datasets of this size or structure," said Gehan. "We're connecting tools that already work and already have users, so the plant science community gets a real, sustainable answer instead of another one-off platform.” 

"Genomics had this moment decades ago, when the field decided that publicly funded sequence data belonged in a shared, searchable home," said Fahlgren. "Image data hasn't had that moment yet. Getting support to build it, at a time when federal science funding is this constrained, says something about how urgent this problem has become." 

PlantCV (“Plant Computer Vision”) has anchored the Danforth Center's role in this project for more than a decade. Developed and maintained by Gehan and Fahlgren's teams since 2014, the open-source toolkit gives researchers a standardized way to extract measurable plant traits, such as growth rate, leaf area, and color, from images collected on equipment ranging from lab microscopes to field-based drones. It has been downloaded more than 62,000 times, has more than 70 contributors worldwide, and has been used in over 175 published studies and 49 graduate theses, with applications that now extend beyond plant science into fields like materials science and animal science. That track record, and the size of the community already relying on the tool, is a central reason NSF is funding the Danforth Center to lead this effort rather than starting from scratch. 

The project will develop tools for packaging PlantCV analysis workflows for reproducibility, build a metadata manifest system in COPO based on the REMBI and MIAPPE community standards, and create a direct pipeline for depositing image data and metadata to the BioImage Archive. The team will also run training workshops through the North American Plant Phenotyping Network and produce public documentation and video tutorials so researchers outside the Danforth Center can adopt the tools. 

“This foundational work lets plant science move faster for everyone," said Giles Oldroyd, PhD, president of the Danforth Center. "Giving researchers a way to dissect and build on phenotypic image data doesn't just serve our own scientists, it serves the global plant science community, including startups and agtech businesses right here in St. Louis.”