flexFitR and exploreHTP: Open‐source software to enable nonlinear modeling of plant growth from remote‐sensing imagery

Abstract In this study, we introduce two complementary R packages: exploreHTP and flexFitR. exploreHTP is a Shiny‐based graphical interface for extracting plot‐level data from remote‐sensing imagery, while flexFitR provides tools for nonlinear model‐fitting and trait derivation from time series. Their application is demonstrated through analysis of a 2022 potato ( Solanum tuberosum L.) breeding trial at the University of Wisconsin. RGB images were collected through eight uncrewed aerial surveys. Three growth‐phase models were explored for ground cover (GC)—linear‐plateau, quadratic‐plateau, and linear‐logistic—coupled with two possible senescence models (none vs. linear decay). Based on penalized goodness‐of‐fit, quadratic‐plateau models were selected for 65% of plots. Plant height (PH) was modeled using squared exponential growth and exponential decay. From the fitted curves, we derived six traits from GC and four from PH. Genotypic correlations with total yield were estimated using a multi‐trait model. Both emergence time and time to maximum GC were negatively correlated with yield, indicating the importance of early canopy establishment. The potential for yield prediction based on growth traits was assessed by data masking. The best secondary trait was average GC, with an accuracy of 0.52, followed by time to maximum GC at 0.50. Combining average GC and maximum height improved accuracy to 0.59. Overall, the flexFitR and exploreHTP packages provide a practical approach to characterize plant growth and support decision‐making in breeding programs.