Abstract:
Pastures are an essential resource for the milk and meat industries, two of the key sectors in the agricultural industry. The main goal of any dairy farmer is to have a productive pasture that can provide nutrients for cattle. One essential factor for quality reduction in any pasture is weed. Dairy farmers spend a tremendous amount of time and budget controlling and destroying weed annually. This study has been defined and developed with a focus on weed detection and weed control. A software system had been designed and developed to monitor pastures for weed control. California thistle was chosen for the study once it was recognised as being the most widespread weed across New Zealand. After a realisation that the core component of any management system is monitoring, a weed-detection model was designed to recognise California thistle in a pasture environment. In the next stage, the model’s accuracy was improved, and a fuzzy inference system was integrated into the model, developing it into a system software for decision making on weed management. The key achievement of this research stage was developing methods to score pastures and generate a 2-D map for weed density and a 2-D map for measuring the quality state of a pasture. No other studies have achieved these supporting outputs. These features in the software system are novel and can help farmers by providing digital insight into their pastures. Finally, it has been thought to deploy the system software in a digital ecosystem by adding collaborative characteristics for sharing weed management knowledge among dairy farmers.
