Machine learning and computer vision for plant health

机器学习和计算机视觉促进植物健康

• 批准号: 517528-2017
• 负责人: Hamarneh, Ghassan
• 金额: $ 1.82万
• 依托单位: Simon Fraser University
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=640425
• 关键词: Machine; learning; computer; vision; plant

Greenhouse plants must be continuously monitored for detecting pests and diseases. It is also important toexamine the crop's fruit production by estimating their numbers and the right time for harvesting. Theseprocesses are currently performed manually via tedious and time consuming visual inspection procedures,which can take weeks for an average size greenhouse. The main focus of this project is on developingcomputational tools for the automated, fast, and accurate interpretation of image data gathered from the plants.In particular, in collaboration with Ecoation, we will develop computer vision and machine learning softwaretools to analyze plant image data and identify important horticultural features and plant health status. Thisproject is essential for the company's development of machine learning and computer vision products tosignificantly reduce the need for human scouting for plant problems due to its precise and timely alertingcapabilities. With the development this project can bring, Ecoation anticipates incremental reductions in croploss due to the use of the Ecoation System in greenhouses. As the result of our early stage detection of cropstress, just in BC, it can save growers $20 million each year.

温室植物必须持续监测，以发现病虫害。通过估计果实的数量和合适的收获时间来检查作物的果实产量也很重要。目前，这些过程都是通过繁琐且耗时的目视检查程序手动完成的，对于一个中等规模的温室来说，这可能需要数周的时间。该项目的主要重点是开发计算工具，用于自动、快速和准确地解释从植物收集的图像数据。特别是，我们将与Ecoation合作开发计算机视觉和机器学习软件工具，以分析植物图像数据并识别重要的园艺特征和植物健康状况。该项目对于公司开发机器学习和计算机视觉产品至关重要，因为它具有精确和及时的警报能力，可以显着减少对人工侦察植物问题的需求。随着该项目带来的发展，Ecoation预计由于在温室中使用Ecoation系统，作物损失将逐步减少。由于我们对作物胁迫的早期检测，仅在不列颠哥伦比亚省，每年就可以为种植者节省2000万美元。