Autonomously-Targeted Weed Management System

萨斯喀彻温省理工学院与专家系统公司合作，正在为农业行业开发一种自主定向的杂草管理系统。杂草往往成簇生长，大片农田中往往没有杂草，这意味着开发一种使用机器学习来识别杂草的神经网络是可能的。因此，除草剂的使用可以自动地专门针对杂草，而不是作物。该项目的第一阶段将涉及在受控环境中将常见的草原杂草和作物播种在一起，并通过高光谱图像对它们进行研究，以开发和训练能够区分它们的神经网络。这将是专家系统的一项极具市场价值的突破，也是萨斯喀彻温理工学院的一项重大科学成就。农业还将受益于生产除草剂使用量大大减少的作物。

In partnership with Expert Systems Inc., Saskatchewan Polytechnic is developing an autonomously-targeted weed management system for the agricultural industry. Weeds tend to grow in clusters and are often absent from large patches of cropland, presenting the possibility of developing a neural network that uses machine learning to identify weeds. Thus, herbicide use can be automated to specifically target weeds, not the crop. The first phase of the project will involve sowing common prairie weeds and crops together in a controlled environment and studying them through hyperspectral imagery in order to develop and train a neural network that can distinguish between them. This will be a highly-marketable breakthrough for Expert Systems and a significant scientific accomplishment for Saskatchewan Polytechnic. The agricultural industry will also benefit by producing crops with vastly reduced herbicide usage.