Drones and Machine Learning Estimate Sorghum Grain Yield in Breeding Trials
· newswise.com · popularity 21/100 · 0 similar stories today · AI & Data Science (Artificial Intelligence)
Read the original at newswise.com →
Summary
Phenotyping remains a bottleneck in sorghum breeding because grain yield is costly and labor-intensive to measure. This study tested whether unmanned aircraft systems (UAS) imagery combined with machine learning (ML) and deep learning (DL) can predict plot-level yield non-destructively. Researchers extracted canopy, vegetation-index and panicle features from red-green-blue (RGB) imagery, using a You Only Look Once version 11 (YOLOv11)-based detector called YOLO-SORG to count panicles. Six models--ridge regression (RR), elastic net (EN), least absolute shrinkage and selection operator (LASSO) regression (LR), support vector regression (SVR), random forest (RF) and XGBoost (XGB)--were compared across three feature sets: panicle traits, canopy traits and both combined. The goal was to identify a practical, repeatable pipeline that could help breeding programs screen material faster and support selection decisions.
Found in: Newswise. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 2.