Open AccessThe Study of Locating Diseased Leaves Based on RPN in Complex Environment
Author(s) -
Yan Guo,
Jin Zhang,
Pei Fang Su,
Guang Hua Hou,
Fang Deng
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1651/1/012089
Subject(s) - boundary (topology) , convolution (computer science) , computer science , artificial intelligence , pooling , feature (linguistics) , frame (networking) , convolutional neural network , position (finance) , object (grammar) , pattern recognition (psychology) , artificial neural network , image (mathematics) , computer vision , mathematics , mathematical analysis , telecommunications , linguistics , philosophy , finance , economics
Plant disease is one of the major factors threatening the plant growth. In this paper, we utilize the region proposed network (RPN) to detect and locate the plant leaf based on the machine deep learning algorithm. Firstly, the original image needs to be input into convolution neural network (CNN). After several convolution and pooling operations, highly condensed image features can be obtained. Secondly, a reference boundary frame for predicting the position of an object can be obtained by sliding nine boundary frames as sliding windows on the feature map. Two neural networks are input into each boundary box to get the classification result and boundary location. Finally, with the help of non-maximum suppression algorithm (NMS), multiple boundary boxes for the same object are eliminated and only the best boundary boxes are retained. Experiments show that RPN algorithm has better performance on locating the diseased leaves in complex environment, thus reducing the influence of disease on agricultural production. At the same time, it is of great significance in economic development, ecological protection, agricultural production and other fields.