Deep Learning to Accelerate Breeding for Crop Disease Resistance

深度学习是人工智能领域一项新颖而令人兴奋的发展，特别适合于解决涉及图像分析的复杂问题。它大体上模仿了人脑编写自己的规则来学习如何对环境中的不同物体进行分类的方式。该项目将为学生提供一个将这项快速发展的技术应用于当前具有重大现实意义的问题的绝佳机会。农业中的一个关键问题是如何在不使用农药的情况下控制作物病害。植物育种者希望推出能抵抗所有重大疾病的新品种，但目前，这需要一名训练有素的人在田间试验中同时评估几种疾病和压力。这是缓慢而困难的，尤其是因为许多病原体产生表面相似的斑点和斑点。这个项目的目的是开发一个深度学习网络，它能够被训练和再训练来识别、区分和量化不同的疾病。这将在田间情况下应用，以提高育种试验中疾病筛选的准确性和严谨性，以选择新的植物品种。

Deep learning is a novel, exciting development in artificial intelligence, particularly suited to solving complex problems involving image analysis. It broadly mimics the way a human brain writes its own rules to learn how to categorise different objects in the environment. This project will give a student an outstanding opportunity to apply this fast-developing technology to a current issue of great practical importance.A key problem in agriculture is how to control crop diseases without pesticides. Plant breeders wish to release new varieties which are resistant to all significant diseases but at present, this requires a highly-trained person to assess several diseases and stresses in a field trial simultaneously. This is slow and difficult, not least because many pathogens produce spots and blotches of a superficially similar appearance.The aim of this project is to develop a deep learning network which is capable of being trained and retrained to identify, distinguish and quantify different diseases. This will be applied in field situations to increase the accuracy and rigour of disease screening in breeding trials to select new plant varieties.