葡萄叶片病害智能检测技术研究进展

打开文本图片集

中图分类号：S661

文献标识码：A

文章编号：1000-4440（2026）04-0842-13

Research progress on intelligent detection technology for grape leaf diseases

WU Jianjun 1，2，3 ， ZHAO Huiyuan 1，2 ， ZHU Yuhua 1，2

College of Information Science and Engineering， Henan University of Technology， Zhengzhou 450001， China; Key Laboratory of Grain Information Processing and Control， Ministry of Education， Henan University of Technology， Zhengzhou 450001， China; Academy of National Food and Strategic Reserves Administration， Beijing 100037， China）

Abstract： Grape leaf diseases pose a severe threat to the sustainable development of the grape industry. Traditional manual diagnosis is plagued by strong subjectivity， low efficiency and poor timeliness， making it difficult to meet the development needs of modern agriculture. In the early stage， intelligent detection of grape leaf diseases was mainly based on traditional machine learning algorithms such as support vector machines， K-nearest neighbors and random forests， with the recognition accuracy gradually rising from 60% to 85%–90%. Convolutional neural networks have effectively addressed the challenge of multi-scale feature extraction through architectural innovations including Inception modules， residual connections and denseconnections， enabling the model recognition accuracy to stabilize at over 95%. The YOLO algorithm has undergone iterative evolution from YOLOv1 to YOLOv1 By integrating strategies such as attention mechanisms， lightweight networks and feature pyramids， it has raised the mean detection accuracy to 90%–98% and simultaneously compressed the model size to the single-digit MB level， achieving the coordinated optimization of accuracy and efficiency.In addition， this paper analyzes the shortcomings of public datasets such as PlantVillage and AI Challenger 2018， and puts forward improvement suggestions for the problems of data standardization and annotation consistency.

Key words： grape; disease; intelligent detection; deep learning; convolutional neural network; YOLO algorithm

葡萄是全球种植规模仅次于柑橘的第二大果树作物，在农业经济体系中占据重要地位[1]。（剩余26153字）