In 2050, the world population is estimated to be over 9 billion people (Godfray et al., 2010). Feeding all these people in a sustainable manner will require adopting new agricultural management practices. In this MSc thesis, we will make use spatio-spectral data acquired from an Unmanned Aerial Vehicle (UAV) to support farmers in their decisions regarding fertilization practices (von Bueren et al., 2015). While fertilization might be needed in some part of the field, other parts might not need fertilizer at all or a different dose. Using data collected by multi- and hyperspectral sensors we want to answer the following research questions: which spatial and temporal patterns linked to fertilization practices can be detected in the data? Which methods from machine learning and data mining fit detection and classification of agricultural management units? To answer these questions, the student will have to cover a wide variety of pattern recognition and machine learning techniques. He/She will also develop a novel recognition system to classify different UAV images of crops based on their spectral features. A UAV was flown over two grasslands fields, containing different types of grasses. During the flights, the cameras mounted on the UAV were used to automatically collect a large number of high-resolution images and spectral samples (Gonzalez et al., 2018). The results of this thesis should help to design a system that learns which parts of the fields deserve a closer look. Moreover, the system should be able to report to the farmer if there are plants that deserve specific attention.