Towards the Computational Assessment of the Conservation Status of a Habitat

We propose methods to automatically assess the conservation status of a habitat. Habitat monitoring is usually performed by botanists and other specialists in their field work, searching for the presence or lack of typical plant species1 and other elements (such as vegetation cover) that might indicate the degradation of a habitat. We present preliminary work that makes use of a robotic platform employed to help botanists in their tasks. Three methods are proposed. First a color segmentation method, to detect the amount of green in a given area, a detection method to automatically detect the presence of a given plant, and finally a classification method used to identify a plant in a single image.