In this project we aim at monitoring the botanical composition of clover-grass fields to allow for targeted fertilization.
Targeted fertilization allows for increasing the yield as well as the quality of the harvested clover-grass, but requires reliable knowledge of the clover fraction in the mixed crops.
Based on previous research, the problem of determining the clover fraction of the harvested dry-matter is split into two parts:
Capture an RGB-image of the clover-grass canopy and semantically segment the image with machine learning techniques.
Combine canopy information with auxiliary data, such as past weather conditions and crop height, to estimate the clover fraction and the expected yield.
Søren Skovsen, Mads Dyrmann, Anders K. Mortensen, Morten S. Laursen, René Gislum, Jørgen Eriksen, Sadaf Farkhani, Henrik Karstoft, Rasmus N. Jørgensen. (2019). The GrassClover Image Dataset for Semantic and Hierarchical Species Understanding in Agriculture. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops
Søren Skovsen, Mads Dyrmann, Jørgen Eriksen, René Gislum, Henrik Karstoft, Rasmus Nyholm Jørgensen. (2018). Predicting Dry Matter Composition of Grass Clover Leys Using Data Simulation and Camera-Based Segmentation of Field Canopies into White Clover, Red Clover, Grass and Weeds. 14th International Conference on Precision Agriculture
Dennis Larsen, Søren Skovsen, Kim Arild Steen, Kevin Grooters, Jørgen Eriksen, Ole Green, Rasmus Nyholm Jørgensen. (2018). Autonomous Mapping of Grass-Clover Ratio Based on Unmanned Aerial Vehicles and Convolutional Neural Networks. 14th International Conference on Precision Agriculture
Søren Skovsen, Mads Dyrmann, Anders Krogh Mortensen, Kim Arild Steen, Ole Green, Jørgen Erikse, René Gislum, Rasmus Nyholm Jørgensen, and Henrik Karstoft. (2017). Estimation of the Botanical Composition of Clover-Grass Leys from RGB Images Using Data Simulation and Fully Convolutional Neural Networks. Sensors, special issue: Sensors in Agriculture