A Novel Approach for Invasive Weeds and Vegetation Surveys Using UAS and Artificial Intelligence

Abstract

Surveillance tasks of weeds and vegetation in arid lands is a complex, difficult and time-consuming task. In this article we present a framework to detect and map invasive grasses, combining UAVs and high-resolution RGB technologies and machine learning for data processing. This approach is illustrated by segmenting Buffel Grass (Cenchrus ciliaris) and Spinifex (Triodia sp.), Segmentation results produced individual detection rates of 97% for buffel grass, 96% for spinifex and 97% for the overall classification task. The algorithm is robust against variations in illumination, occlusion, object rotation and density of vegetation.