基于YOLO V3的葡萄病害人工智能识别系统

Intelligent identification system of grape diseases based on YOLO V3

葡萄在生长过程中会感染各种病害,葡萄病害的高效识别是防治葡萄病害的关键。本文提出了一个基于YOLO V3的葡萄病害智能识别系统,由微信小程序、云服务器和葡萄病害识别模型构成。其中的葡萄病害识别模型以植物病虫害生物学国家重点实验室提供的2 566张原始葡萄病害图片为基础,构建了32 871张葡萄病害图片数据集,采用改进的YOLO V3训练得到。本系统能对手机保存的或现场拍摄的自然条件下的12类葡萄病害图像进行识别,准确率达98.60%。识别结果、病害特征、发病原因、病害地理分布和防治建议可立刻反馈至用户。本系统不但识别率高,而且涵盖了目前大多数常见葡萄病害种类,可作为辅助果农、消费者和相关科研人员甄别葡萄病害的智能工具。

Grape is infected with various diseases during its growth process, and efficient identification is the key to the prevention and control of grape diseases. In this study, an intelligent grape disease recognition system was proposed, which was composed of WeChat mini program, cloud server and grape disease identification model. The grape disease identification model was based on 2 566 original grape disease pictures provided by the State Key Laboratory for Biology of Plant Diseases and Insect Pests. A data set composed of 32 871 grape disease images was constructed and trained by improved YOLO V3. The system could identify 12 kinds of grape diseases, with an accuracy of 98.60%. The identification results, characteristics, causes, geographical distribution and control suggestions of grape diseases could be immediately fed back to the user. This system not only had a high recognition rate, but also covered most of the common grape diseases available at present, which can be used as an intelligent tool to assist growers, consumers and researchers for the identification of grape diseases.