Machine Learning-based Brassica Napus L. Study On The Disease Degree Method Of The Classification Of Clubroot

Rape root swelling disease will affect the quality of rape roots.It is particularly important to predict the disease before it causes serious damage to plants.Therefore,in order to overcome the difficulty of insufficient accuracy caused by too large deviation of Brassica napus root swelling disease detected and identified manually and completely relying on personal subjective judgment,this paper proposes a method of classifying root swelling disease by using an efficient expression feature sparse structure model through image processing and analysis and machine learning technology.Firstly,through the preprocessing of the collected original Brassica napus pictures,the rape pictures are cut and segmented in turn to obtain a single root image with high quality.Then,the segmented root image is binarized and denoised.In order to better extract the characteristics of rape root image,a rape root eigenvalue extraction system is designed based on the GUI platform of MATLAB software to extract all effective eigenvalues of rape root area.After grouping,by combining the features extracted from these images and introducing the input of transfer learning model,a classification model of rape root swelling disease degree based on perception is proposed,and satisfactory results are obtained.In this paper,through the analysis of the disease degree of Brassica napus root image,the traditional neural network algorithm is improved,and compared with the traditional BP neural network recognition method,the calculation performance of the neural network is improved,so that it can greatly enhance the recognition of Brassica napus root swelling disease.The accuracy of identification has been improved by about 10%,which can more effectively improve the effectiveness of identification of diseases such as rape production root swelling,and more meet the needs of modern agricultural production practice.