MSc Thesis Defense by Jesper Lundsgaard


Drone image analysis of cereal fields


Drone technology has many great uses in agriculture, which are being actively explored and prioritised. One of those uses is monitoring of weed in cereal fields, which can helps farmers keep down weed while also saving the environment from unnecessary pollution, by only spreading weed killer where there is any weed. This thesis builds on previous research, which presented a way to detect weed and combine the detections from individual drone images into a global coordinate system. 

The thesis shows that results of the comparable quality can be achieved using much simpler techniques, that exploit the fact, that the need for precision is  counted in meters, not centimeters. This is shown by presenting a method where  weed detection is performed on individual images and those detections are mapped using some different homographies. By showing this, the thesis makes a point of choosing the right tool for the job and not wasting excessive amounts of resources on something, that can be solved quickly.

Supervisor: Søren Olsen

Censor: Søren Overgaard