Plant Wilting Estimation And Field-Based Plot Extraction

Plant phenotyping is the process of characterization and quantification of physical traits
of plants such as height, leaf area, biomass, wilting degree, or flowering time. Many plants
become limp or droop through heat, loss of water, or disease. This is also known as wilting.
In this thesis, we propose multiple quantifiable wilting metrics that will be useful in studying
bacterial wilt and identifying resistance genes. In order to obtain the wilting metrics, we use
machine learning methods to identify the center of the stem. We also propose a fast ground
truthing method to speed up training data generation. We test our metrics on both tomato
plants and soybean plants with wilting caused by either bacteria or drought. We successfully
demonstrated that our metrics are effective at estimating wilting in plants.

Field experiments often comprise thousands of plants. For many Unmanned Aerial Vehi-
cles (UAVs) image-based plant phenotyping analyses, we need to examine smaller groups of
plants known as ”plots”. We propose a method to extract plots from images acquired from
UAVs. In addition, we proposed a system that will allow us to combine our plot extraction
results with field data such as plant ID, plant genotype, and experiment type provided by
the planters. We also developed a method to generate synthetic plant center location data.