基于深度学习的中药材病虫害图像识别系统设计

Key techniques of image processing and in-depth learning for identification of Chinese medicinal materials

针对传统中药材粉末显微鉴定图像识别精度低的问题,提出设计一个基于深度学习的中药材病虫害图像识别系统。首先,该系统以SE-MobileNetV2作为基础网络模型;然后基于基础网络模型引用Momentum优化算法与ReLU6激活函数,通过这两种算法对基础网络模型进行改进和训练;最后将训练好的网络模型进行中药材粉末显微鉴定图像准确识别。实验结果表明,在选取的8种中药材粉末显微鉴定图像中,改进的网络对4种中药材粉末显微鉴定图像的识别准确率为100%,其余4种的识别准确率均高于93%,总体的识别准确率为97.5%。由此证明,本系统具有较高的识别性能,很大程度上提升了中药材粉末显微鉴定图像识别的精度和效果,对中药材病虫害预防与检测具有积极意义。

Aiming at the problem of low accuracy of traditional TCM,a new image recognition system based on deep learning.First,the system uses SE-MobileNetV2 as the basic network model;then the basic network model uses the Momentum optimization algorithm and ReLU 6 activation function to improve and train the basic network model;and finally the trained network model can ac-curately identify the microscopic image of TCM powder.The experimental results show that the recognition accuracy of the improved network of the 8 kinds of Chinese herbal medicine powder is100%,the other 4 recognition accuracy is higher than 93%,and the over-all recognition accuracy is 97.5%.This proves that the system has high recognition performance,which greatly improves the accuracy and effect of powder microscopic identification of Chinese medicinal materials,and has positive significance for the disease and insectpest prevention and detection of Chinese medicinal materials.