I-Corps: Translation Potential of a Smartphone-Based Crop Disease Detection Application

这个i-Corps项目利用经验学习和对行业生态系统的第一手调查来评估该技术的翻译潜力。该项目专注于开发一款用于农业作物病害检测的智能手机应用程序。及早发现病虫害有助于遏制疾病的传播。虽然作物病害可能对作物产量产生重大影响，导致农民经济损失，并可能影响粮食安全，但目前还没有可靠的基于计算机视觉的解决方案来帮助研究人员和农民在田间实现可靠的疾病评分。这项技术是一种用于植物病害检测的移动应用程序，它可以自动在田间识别作物病虫害，并超越了人工田间侦察可能实现的目标。这项技术的使用可能会提高作物产量，并使农业生产中的管理和决策更加有效。该技术是基于使用智能手机应用程序进行农业作物病害检测的先前开发的。被病虫害感染的植物通常在叶、茎、花或果实上有记号或斑点。该技术旨在部署在云中进行实时疾病检测，并使用先进、轻量级的深度学习模型和直接从农业领域收集的大图像集来覆盖所有场景，以实现准确的疾病检测结果。此外，该技术旨在根据用户的需求以多种方式提供检测结果，包括疾病评分和图像，以及用于可视化和测量病变区域大小或计数的分割图像。该应用程序可用于收集适当的作物图像并将其记录到云中。与传统的人工计分方法相比，这种应用程序可以显著提高虫害和疾病检测效率，同时减少人工使用并消除人类专业人员的专业知识。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This I-Corps project utilizes experiential learning coupled with first-hand investigation of the industry ecosystem to assess the translation potential of the technology. The project focuses on the development of a smartphone application for agricultural crop disease detection. Early detection of pests and diseases helps contain the spread of the disease. While crop diseases can have a significant impact on crop yields, leading to economic losses for farmers and potentially affecting food security, there are no reliable computer vision-based solutions to help researchers and farmers achieve reliable disease scoring in the field. This technology is a mobile application for plant disease detection that automates crop pest and disease recognition in the field and goes beyond what may be achieved with manual field scouting. The use of this technology may improve crop yield and empower more efficient management and decision making in agricultural production.The technology is based on the prior development of agricultural crop disease detection using a smartphone application. Plants infected by pests or diseases typically have marks or lesions on the leaves, stems, flowers, or fruits. The technology is designed to be deployed in the cloud for real-time disease detection and uses advanced, lightweight deep learning models and large image sets collected directly from agricultural fields covering all scenarios to achieve accurate disease detection results. In addition, the technology is designed to provide detection results in multiple modalities based on users’ needs including disease score and images with segmentation for visualization and measurements of the size or count of diseased areas. The application may be used to collect and record the appropriate crop images to the cloud. This application may significantly speed up pest and disease detection efficiency while reducing labor use and eliminating expertise from human professionals as compared to the traditional scores-in-notes manual method.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.