Accurate information about a crop is needed to apply a precision treatment to it as well as to perform other agricultural tasks. To achieve site-specific management of weeds, the first and most important step is the location and density estimation of weeds. In this context, the development of computer vision methods for real-time weed detection can be highly useful in the construction of fully automatic devices for weed control. This paper presents a visual method that discriminates between crop rows and weeds in wide-row crops, working in real time with the images acquired from a conventional camera on board a tractor and under uncontrolled lighting and movement conditions.

This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no 245986

Peer Reviewed