Towards Automated UAV Image Processing Workflows in Precision Viticulture: Challenges and Benefits

Abstract

Remote sensing is a tool based on acquiring images of objects at a distance. The distance to the object is a determining factor, conditioning the strength of the analysis, and it can be reduced with lower flights and longer flight missions. Still, there is a physical limit depending on the type of platform used. In this way, Unmanned Aerial Vehicles (UAVs) allow images to be obtained with a higher spatial resolution, as they can fly only a few metres above the ground. Therefore, they have become a very useful platform as they have democratised field imaging technology, offering greater versatility in obtaining spectral information due to their small size and significantly reduced costs compared to traditional aerial platforms. In this way, several authors have conducted comparative studies on the suitability of different platforms for remote sensing data collection in viticulture; however, there is a lack of information about the image processing workflows. In the framework of the FlexiGroBots project, methodologies for analysing the spatial variability of different factors affecting the crop are being explored, and the automation of processes is a key factor that has proven their applicability. In this article, the suitability of carrying out two different workflows for image processing is analysed: i) a manual workflow, which involves specific steps with human intervention and considers software traditionally used by pilots and remote sensing technicians, and ii) an automated workflow in Python, which allows image processing to be carried out with little or no human intervention.