GrowPro: a flexible and high-resolution imaging tool for high-throughput plant phenotyping in the field

Abstract

Various remote sensing platforms are being used in agriculture research, but their capability and exibility may not be enough for carrying out e cient computer analysis. Low-quality data can a ect the accuracy of methods and techniques used for high-throughput phenotyping in plant breeding. Despite the fact that there are many available remote sensing tools, researchers are seeking new platforms with superior features and performance over already proven systems. In this thesis, we describe a novel approach for remote sensing for eld-based phenotyping called the GrowPro. This thesis presents an overview of related work and commonly available approaches to remote sensing, description of the designed system, data gathering procedure, post-processing pipeline and best practice for capturing plants. We show the ease of use of the GrowPro, simplicity of data gathering and qualify the accuracy of the stitching process. We proved that by using this novel approach, high-resolution RGB stitched images of regions of interests (e.g., an individual plot or range of plots) can be obtained. This method appeared to be stable over time, di erent trials and weather conditions. Examples of RGB stitched images of a variety of crops at various stages of growth have been described.