福建电脑2025，Vol.41Issue(5)：18-22,5.DOI:10.16707/j.cnki.fjpc.2025.05.004
改进ResNet50的番茄叶片病害识别方法
Improved Method of Tomato Leaf Disease Identification by ResNet50
张晨晨1
作者信息
- 1. 周口文理职业学院信息工程学院 河南 周口 466000
- 折叠
摘要
Abstract
In response to the problems of low accuracy and poor classification performance of traditional convolutional neural network ResNet50 for tomato leaf disease recognition,this paper proposes a tomato leaf disease recognition method that integrates transfer learning and mixed attention mechanism.Firstly,the ResNet50 network pre trained on the ImageNet dataset is utilized to transfer and fine tune the model parameters;Then,in the final convolution stage of the ResNet50 model,a mixed attention mechanism module CSS is introduced,which can enhance the network's ability to extract local detailed features of tomato diseases,help prevent overfitting,and improve the model's generalization ability.The experimental results show that the improved ResNet50 TL-CSS network model has improved accuracy by 1.88%compared to the original model,reaching 97.31%.This method can accurately extract the characteristics of tomato leaf diseases and efficiently identify tomato leaf virus damage.关键词
番茄叶片/病害识别/卷积神经网络/混合注意力机制Key words
Tomato Leaves/Disease Identification/Convolutional Neural Network/Mixed Attention Mechanism分类
计算机与自动化引用本文复制引用
张晨晨..改进ResNet50的番茄叶片病害识别方法[J].福建电脑,2025,41(5):18-22,5.
