Project Articles

Default Latest Most Read

Please wait a minute... For Selected: Download Citations EndNote Ris BibTeX Toggle Thumbnails Select Image Recognition of Wheat Leaf Diseases Based on Lightweight Convolutional Neural Network and Transfer Learning FENG Xiao, LI Dandan, WANG Wenjun, ZHENG Guoqing, LIU Haijiao, SUN Yongsheng, LIANG Shan, YANG Ying, ZANG Hecang, ZHANG Hui, Journal of Henan Agricultural Sciences    2021, 50 (4): 174-180.   DOI: 10.15933/j.cnki.1004-3268.2021.04.023 Abstract （321）      PDF （3277KB）（433）       Knowledge map    Save In order to realize the convenient recognition of wheat leaf disease image on mobile terminal,image recognition model of wheat leaf disease was established based on lightweight convolutional neural network (CNN) and transfer learning.Firstly,a sample data set of 224 pixel×224 pixel was established,which was composed of images of wheat powdery mildew,stripe rust and leaf rust. Secondly,the deep learning framework Tensorflow 2.0 was used to build the wheat leaf disease image recognition model based on MobileNetV2, and the trained parameters on ImageNet data set were used as the initial parameters of the model. Finally,the effects of transfer learning method,sample size,adding Dropout layer before global average pooling and initial learning rate setting on the model performance were analyzed.The results showed that the average recognition accuracy of the images of three wheat diseases was as high as 99.96% by setting all layers of the model as trainable,selecting suitable data enhancement method to increase the sample size, adding Dropout layer before global average pooling and setting the initial learning rate of 0.000 01.Based on MobileNetV2 and transfer learning, a wheat leaf disease image recognition model with high recognition accuracy and strong generalization ability could be constructed for mobile terminal. Related Articles | Metrics Select Stem and Leaf Segmentation of Maize Plant Image Based on Skeleton Extraction and Binary Tree Analysis ZHANG Weizheng, LI Xuguang, WAN Hanwen, LI Canlin, ZHANG Weiwei, JIN Baohua, LIU Yan Journal of Henan Agricultural Sciences    2020, 49 (9): 166-172.   DOI: 10.15933/j.cnki.1004-3268.2020.09.021 Abstract （266）      PDF （2767KB）（448）       Knowledge map    Save Segmentation of stems and leaves of corn plant images could provide parameters and basis for subsequent studies such as phenotypic analysis and plant response to environmental stress.Maize plant images were selected from the plant phenotypic dataset of the University of Nebraska-Lincoln,and the maize plants image were tailored and binarized automatically.The skeleton model of the plant was established by the skeleton refinement algorithm,and the threshold was set to remove the burrs in the skeleton;then the end points and branch points on the skeleton were detected,the skeleton model was analyzed by the binary tree,and the nodes at each level and the individual leaves were finally established,thereby realized the segmentation of the stems and leaves.The results showed that the established stem and leaf segmentation algorithm of maize plants has the advantages of fast processing speed and easy understanding,and provides support for the phenotypic analysis and breeding of plants such as maize. Related Articles | Metrics Select Identification of Diseases and Insect Pests of Malu Grape Based on CSJMM⁃AS⁃GAC WANG Xingwang, ZHENG Hanyuan, WANG Suqing Journal of Henan Agricultural Sciences    2022, 51 (6): 154-163.   DOI: 10.15933/j.cnki.1004-3268.2022.06.017 Abstract （206）      PDF （4107KB）（172）       Knowledge map    Save In order to improve the identification accuracy of diseases and insect pests of Malu grape and effectively carry out the prevention and control of diseases and insect pests of Malu grape，this paper improved the geodesic active contour model（GAC），by introducing the dynamic coefficient function，the boundary region and non⁃boundary region of Malu grape pest image were accurately divided，so as to accurately segment the fuzzy and concave boundary of the pest image. A precise segmented geodesic active contour model（AS⁃GAC） was proposed and established. Next，in order to overcome the error caused by insufficient training samples in complex background，the Core loss function was proposed，and the Core⁃Softmax joint supervision mechanism（CSJMM） was established. Thus，a precise segmented geodesic active contour model based on CSJMM（CSJMM⁃AS⁃GAC）was established. The results showed that the initial accuracy of CSJMM⁃AS⁃GAC training set was 65.46%，the accuracy of validation set was 95.67%，the accuracy of test set was 93.95%，the Kappa coefficient of CSJMM⁃AS⁃GAC reached 0.913 8，the recall rate reached 89.21%，and the average accuracy of CSJMM⁃AS⁃GAC for identifying the diseases and pests of Malu grape reached 94. 06%. The overall performance，recognition accuracy and recall rate of CSJMM⁃AS⁃GAC are better than the commonly used disease and pest identification models. Table and Figures | Reference | Related Articles | Metrics Select Intelligent Acquisition of Rice Disease Images Based on Python Crawler and Feature Matching YANG Tianle, QIAN Yinsen, WU Wei, SUN Chengming, LIU Tao Journal of Henan Agricultural Sciences    2020, 49 (12): 159-163.   DOI: 10.15933/j.cnki.1004-3268.2020.12.023 Abstract （192）      PDF （1256KB）（324）       Knowledge map    Save For timely diagnose and prevent the rice diseases,computer technology and image processing technology were used for disesae diagnosis. Python crawler technology was used to compile image crawler programs based on rice disease keywords.The feature matching of Matlab image was used to filter the image set to improve the accuracy of image collection.The results showed that the extraction accuracy of rice disease images obtained only by Python crawler technology was higher than 50. 00%,except bipolaris oryzae.Among them,the extraction effect of gibberellic disease was the best,with an accuracy rate of 72.7%.The false detection rate of images after the feature matching screening was below 6.00%,which not only improved the accuracy of data collection,but also showed that rice diseases image acquisition through the intelligent method was feasible. Related Articles | Metrics Select Wheat Seedling Identification Based on K-means and Harris Corner Detection XU Xin, LI Haiyang, FENG Yangyang, MA Xinming, SHEN Shuaijie, QIAO Xinyu Journal of Henan Agricultural Sciences    2020, 49 (12): 164-171.   DOI: 10.15933/j.cnki.1004-3268.2020.12.024 Abstract （172）      PDF （4916KB）（353）       Knowledge map    Save To solve the problem of low efficiency and time-consuming for the field investigation methods of wheat seedling count,we systematically studied the influence of the images acquired from the 1—4 leaf stage of wheat at different shooting angles of mobile devices on the number of wheat seedlings identified.Markers were set to locate and segment the two-row 1-meter area to be identified,and image processing technology was used to precisely cut and correct the target area. On this basis,the effects of four image segmentation methods on wheat seedling image segmentation were compared,and the image was automatically segmented into different connected regions using image segmentation algorithm,then the connected domain was used extraced for inner cavity filling and adhesion of wheat leaf apex angular point block processing,and the Harris corner detection algorithm was adopted to process the wheat seedling stem endpoint for identification.According to the one-to-one relationship between wheat and wheat seedling stem,the basic number of wheat seedlings was calculated. The experimental results showed that different mobile devices did not affect the wheat seedling identification accuracy,the best shooting angle of wheat seedling image was 45°,and K-means clustering segmentation algorithm had the best segmentation effect on basic wheat seedlings. At different shooting times, with the increase of wheat leaves, the identification accuracy decreased gradually,and the identification accuracy was more than 0.97 in the 1—2 leaf stage, R 2=0.99.The identification accuracy in the 3—4 leaf stage was more than 0.95, R 2= 0.93,indicating that the application of image segmentation based on K-means fast segmentation combined with Harris corner detection method for quick,accurate and intelligent monitoring and identification was feasible. Related Articles | Metrics Select The Recognition of Early Tobacco Disease Based on FA-SVM Technology ZHANG Hongtao, ZHU Yang, TAN Lian, XU Shuaitao, LIU Jia’nan Journal of Henan Agricultural Sciences    2020, 49 (8): 156-161.   DOI: 10.15933/j.cnki.1004-3268.2020.08.020 Abstract （162）      PDF （3086KB）（308）       Knowledge map    Save For the accurate identification of tobacco disease and providing scientific basis for formulating control methods.A recognition method based on(FA-SVM)technology was proposed.The tobacco brown spot and frog eye disease were selected as the research object.The leaves of tobacco plants with two diseases were photographed by visible light.The segmentation and morphology methods were used to acquire disease spot images.Thirty-two features of spot were extracted to construct the original feature space,including color features,morphological features and texture features.The ant colony optimization(ACO) algorithm was used to extract the partial features build of the optimal feature space by the fitness function.The thirteen features were determined and the max fitness value was 95.68.The firefly algorithm(FA) was used to optimize the penalty factor(c)and the kernel function parameter(g)of support vector machine(SVM).The recognition accuracy of the classification model reached 96% when c=94.12,g=2.43.The results showed that the identification of tobacco disease is feasible based on FA-SVM technology. Related Articles | Metrics