Unsupervised detection of crop rows from images taken by an unmanned aerial vehicle

This study introduces a methodology for processing high-resolution spatial images captured by unmanned aerial vehicles (DAVs) with the goal of detecting crop rows in peri- urban horticultural fields. The image processing approach is unsupervised and solely relies on open-source tools. This ensures that the procedure can be easily generalized and adapted to other datasets with similar features. The article details the sources and attributes of the data used, and outlines the algorithms and concepts involved in the processing journey, culminating in the creation of the final product. The process encompasses the preprocessing of images, field segmentation into individual plots, the rectification and cropping of each plot, and the subsequent detection of crop rows within these plots. The presented methodology holds promise in automating and enhancing the accuracy of crop row detection in horticultural settings, thereby offering invaluable insights for agricultural planning and decision-making.