Learning to Detect Vegetation Using Computer Vision and Low-cost Cameras

GASTON ARAGUAS; JAVIER REDOLFI; ANA GEORGINA FLESIA; SERGIO FELISSIA

Buenos Aires

Conferencia; IEEE International Conference on Industrial Technology; 2020

ITBA

A problem of current agriculture is the large amountof agrochemicals used to boost production due to their costand the environmental pollution they cause. A partial solutionto this problem consists in developing selective spraying tech-niques through the measurement of a green index that allowsthe selection of the precise amount of pesticide to be appliedaccording to the specific conditions of each part of the field.Some of the problems of the existing systems are the inabilityto discriminate between types of vegetation and to pinpoint itslocation, since they only detect general patches of vegetation. Inthis work, we introduce a system prototype capable of measuringthe presence of vegetation in an area using low-cost devicescombined with current computer vision techniques. The systemallows to generate a mask with the presence of vegetation ina certain area and it is also capable of distinguishing betweendifferent materials unlike current methods, which only allow todistinguish between green and non-green areas. The presentedmethod opens the door to future research which can allowdistinguishing between crops and weeds to make an even moreselective application. The output of the system can be used alsoto design another type of weeding method that is not based onthe application of agrochemicals.