J. Weyler: Joint Plant Instance Detection and Leaf Count Estimation for In-Field Plant Phenotyping

Jan Weyler is a Ph.D. student at the Photogrammetry Lab at the University of Bonn since February 2019. He received his master’s degree at the Institute of Geodesy and Geoinformation in 2019. Before starting his Ph.D., he was working as a student assistant for the EU funded project FLOURISH at the Institute of Photogrammetry.

Precision management of agricultural fields as well as plant breeding play important roles in keeping yields high and to provide food, feed, and fiber for our society. A central element in breeding trials but also for targeted management actions is to analyze the growth state of individual plants objectively and at large scale. In this paper, we address the problem of analyzing crops in real agricultural fields based on camera data and to derive information about the plant development.

We propose a novel single-stage object detection approach that can localize crops and weeds in the field. At the same time, it detects plant-specific leaf keypoints intending to estimate leaf count at a plant level, which is a key trait for classifying the growth stage using the BCCH index. We implemented and thoroughly test our approach on real field data. It can perform the required detections and shows superior performance with respect to a state-of-the-art two- stage approach based on Mask R-CNN.