Plot-Level Maize Early Stage Stand Counting and Spacing Detection Using Advanced Deep Learning Algorithms Based on UAV Imagery

Phenotyping is one of the most important processes in modern breeding, especially for maize, which an crop food, feeds, and industrial uses. Breeders invest considerable time identifying genotypes with high productivity stress tolerance. Plant spacing plays a critical role determining yield crops production settings to provide useful management information. In this study, we propose automated solution using unmanned aerial vehicle (UAV) imagery deep learning algorithms accurate stand counting plant-level variabilities (PSV) order facilitate breeders’ decision making. A high-resolution UAV was used train three models, namely, YOLOv5, YOLOX, YOLOR, both maize PSV detection. The results indicate that after optimizing non-maximum suppression (NMS) intersection union (IoU) threshold, YOLOv5 obtained best accuracy, coefficient determination (R2) 0.936 mean absolute error (MAE) 1.958. Furthermore, YOLOX model subsequently achieved F1-score value 0.896 This study shows promising accuracy reliability processed automating evaluation its potential be implemented further into real-time breeding