The goal of the US Government-funded DEEPER project is an integrated platform of phenomic, genomic, and in silico technologies generating maize lines with deeper roots. Deeper-rooted maize would i) increase carbon sequestration from the atmosphere into deep soil strata, ii) improve crop and biofuel productivity under drought and iii) improve N capture by crops.
CVL’s role in the project is to develop advanced image analysis and deep learning approaches to automatically and rapidly extract anatomical and cell wall composition metrics from the 3D images provided by LAT2.0, an improved laser ablation tomography imaging system also being developed within DEEPER. LAT image analysis will be framed as an instance segmentation task in which the identification of image sub-areas likely to arise from different objects is combined with the classification of those regions by object type. A graph-based description of root anatomy will be constructed and custom-built visualization tools will allow the user to query and browse the resulting rich 3D data structures.
DEEPER is supported by a grant from ARPA-e, the research arm of the US Dept. of Energy.
Pennsylvania State University
University of Wisconsin
University of Georgia
University of Nottingham (Malcolm Bennett, Sacha Mooney, Tony Pridmore)