We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于多尺度和注意力机制的番茄病害识别方法.
- Authors
张宁; 吴华瑞; 韩笑; 缪祎晟
- Abstract
The timely detection and management of eggplant diseases can help increase the yield and quality of tomatoes, and increase the economic benefits of farmers. The use of Internet of Things and artificial intelligence can effectively detect tomato diseases without damage. This research proposes an improved AT-InceptionV3 (Attention-InceptionV3). ) Neural network tomato leaf disease detection model, the network uses InceptionV3 as the backbone network, combined with multi-scale convolution and attention mechanism CBAM (convolu-tional block attention module, CBAM) module, which enhances the expression of disease information and suppresses irrelevant information interference ; At the same time, migration learning is introduced to prevent overfitting when the sample data volume is small. In order to evaluate the effectiveness of the optimization model, an experimental simulation test was carried out on the Plant Village public tomato disease data set. The improved model was tested in the testing phase Tomato healthy leaves, bacterial spot disease, late blight, leaf mold and yellow squishy 5 kinds of tomato common leaf image classification accuracy reached 98.4%, the optimization effect is significant. In order to further verify the universality of this method in different IoT The experiment compared the classification effect of the model on disease images with different resolutions. The results show that the partial loss of image accuracy will not reduce the accuracy of disease classification. This model can provide an important basis for decision-making and judgment of the tomato greenhouse intelligent network.
- Publication
Acta Agriculturae Zhejiangensis, 2021, Vol 33, Issue 7, p1329
- ISSN
1004-1524
- Publication type
Article
- DOI
10.3969/j.issn.1004-1524.2021.07.19