Parker Travis A, Celebioglu Burcu, Watson Mark, Gepts Paul
Department of Plant Sciences, University of California Davis, Davis, CA, United States.
Department of Animal Science, University of California Davis, Davis, CA, United States.
Front Plant Sci. 2025 Aug 12;16:1630162. doi: 10.3389/fpls.2025.1630162. eCollection 2025.
Technological advances have made drones (UAVs) increasingly important tools for the collection of trait data in plant science. Many costs for the analysis of plant populations have dropped precipitously in recent decades, particularly for genetic sequencing. Similarly, hardware advances have made it increasingly simple and practical to capture drone imagery of plant populations. However, converting this imagery into high-precision and high-throughput tabular data has become a major bottleneck in plant science. Here, we describe high-throughput phenotyping methods for the analysis of numerous plant traits based on imagery from diverse sensor types. Methods can be flexibly combined to extract data related to canopy temperature, area, height, volume, vegetation indices, and summary statistics derived from complex segmentations and classifications including using methods based on artificial intelligence (AI), computer vision, and machine learning. We then describe educational and training resources for these methods, including a web page (PlantScienceDroneMethods.github.io) and an educational YouTube channel (https://www.youtube.com/@travisparkerplantscience) with step-by-step protocols, example data, and example scripts for the whole drone data processing pipeline. These resources facilitate the extraction of high-throughput and high-precision phenomic data, removing barriers to the phenomic analysis of large plant populations.
技术进步使无人机（无人驾驶飞行器）日益成为植物科学中收集性状数据的重要工具。近几十年来，植物群体分析的许多成本急剧下降，尤其是基因测序成本。同样，硬件的进步使得获取植物群体的无人机图像变得越来越简单和实用。然而，将这些图像转化为高精度、高通量的表格数据已成为植物科学中的一个主要瓶颈。在此，我们描述了基于来自不同传感器类型图像的高通量表型分析方法，用于分析众多植物性状。这些方法可以灵活组合，以提取与冠层温度、面积、高度、体积、植被指数以及从复杂分割和分类中得出的汇总统计数据相关的数据，包括使用基于人工智能（AI）、计算机视觉和机器学习的方法。然后，我们介绍了这些方法的教育和培训资源，包括一个网页（PlantScienceDroneMethods.github.io）和一个教育性的YouTube频道（https://www.youtube.com/@travisparkerplantscience），其中包含整个无人机数据处理流程的分步协议、示例数据和示例脚本。这些资源有助于提取高通量和高精度的表型组学数据，消除了对大量植物群体进行表型组学分析的障碍。
