In this context, the ideal setup would be to have at our disposal
a tool providing real-time information on weed infestation
and location at each point, which could then be used to activate a
sprayer in response. This work presents an approach aimed at precisely
extracting weed, crop and soil covers from each video frame,
in real-time. The main challenge in terms of image analysis is to
achieve an appropriate discrimination among weeds, crops and soil
under varying conditions of lighting, soil background texture, crop
damage and weed infestation, within the strong processing time
limitations imposed by the spraying operation.