Unoccupied aerial vehicle (UAV)-based high-throughput phenotyping provides scalable and cost-effective access to phenotypic information for crop improvement, yet its application in minor crops such as lentil remains limited. This study applied UAV-derived canopy traits and growth modeling to a nested association mapping population developed from CDC Redberry crossed with 32 diverse founder lines. UAV imagery collected across four site-years was used to captured canopy height, crop area, and crop volume per plot basis at multiple time points. Logistic growth models were fitted to derive crop growth parameters, maximum canopy size, growth rate, and cumulative growth anchored to phenological stages. These static and time-series traits were evaluated for seed yield prediction using partial least squares regression with a 70:20:10 data split and 10-fold cross-validation. Static traits such as maximum crop volume and maximum crop area were consistently associated with yield. Dynamic trait–based models improved prediction accuracy and identified the swollen pod stage (R5–R6) as the most informative forecasting window. External validation using an independent trial confirmed the generalizability of the approach. This study presents a UAV phenotyping framework that supports crop growth dissection and early yield prediction, and downstream trait discovery in lentil.