Vision Based Crop Row Detection for Low Cost UAV Imagery in Organic Agriculture

Abstract

Unmanned aerial vehicles (UAVs) and autonomous smart farming robots are a great solution to reduce the use of chemical or synthetic pesticides. One of the critical aspects of these systems are the requirement to follow the crop rows without damaging the plants or the dam. In this paper, a system was developed which enables a semi-autonomous flight by means of crop row detection in organic agriculture. The system consists of a Raspberry Pi microcontroller with a camera and uses a number of image processing methods to detect plant rows. The evaluation of images of different crop row fields and of crop row models showed that all different plant rows were reliably detected. A runtime evaluation showed that the computing power of the microcontroller for the application is limited. Optimizations such as reducing the image size and adapting calculation-intensive methods resulted in a frame rate of 8 to 12 FPS. This processing time is sufficient for a slow UAV flight.