WP Monitoring delivers up-to-date methods for detecting and mapping perennial weeds.
Perennial weeds grow in patches, which complicate weed control experiments, population dynamic studies and the estimation of their competitive abilities. The main reason is that perennial weeds traditionally have been manually monitored, which is practically impossible due to high labor costs. Furthermore, the patchy distribution of perennial weeds calls for site-specific weed management, which requires cost-effective weed monitoring.
Main emphasis in the work package will be on unmanned aerial vehicle (UAV) imagery because UAV data collection is cost-effective and new weed detection methods based on aerial images show promising potentials. Success criteria will be detection accuracy, capacity, cost-effectiveness and ease of use. To benefit from the fast-developing research of image analysis (e.g. convolutional neural networks), annotated images will be shared to make validation and benchmarking possible for computer scientists.
Jesper Rasmussen Jesper Rasmussen Jesper Rasmussen
