基于Caffe卷积神经网络的大豆病害检测系统

以常见的大豆病害图片为样本,研究分析了大豆的叶斑病、花叶病、霜霉病和灰斑病,并利用卷积神经网络技术设计了针对大豆的病害检测系统。通过对病害图片的二值化和轮廓分割等预处理来获得神经网络模型的训练集,并在此基础上对模型进行了多方面的优化,利用Caffe框架对优化后的网络模型进行了识别率等方面的实验验证。此外,为提高模型使用的便捷性,本实验使用了Qt软件为该系统设计了人机交互界面,从而进一步实现了数据可视化。

Soybean disease detection system based on convolutional neural network under Caffe framework

The diseases such as leaf spot, mosaic, downy mildew and gray spot of soybean were analysed, and then a soybean disease identification system based on convolutional neural network was proposed. The training set of the neural network model was obtained by the pretreatments including binarization of disease images and extraction of target regions, moreover, the accuracy of the model was improved, and the model and related parameters were simulated under the Caffe framework. Furthermore, in order to improve the ease and reliability of the system in use, the human-computer interaction interface was designed by using Qt software. The data visualization was further realized.