Maize leaf disease identification using deep transfer convolutional neural networks

Corn diseases are one of the significant constraints to high–quality corn production, and accurate identification of corn diseases is of great importance for precise disease control. Corn anthracnose and brown spot are typical diseases of corn, and the early symptoms of the two diseases are similar, which can be easily misidentified by the naked eye. In this paper, to address the above problems, a three–dimensional–two–dimensional (3D–2D) hybrid convolutional neural network (CNN) model combining a band selection module is proposed based on hyperspectral image data, which combines band selection, attention mechanism, spatial–spectral feature extraction, and classification into a unified optimization process. The model first inputs hyperspectral images to both the band selection module and the attention mechanism module and then sums the outputs of the two modules as inputs to a 3D–2D hybrid CNN, resulting in a Y–shaped architecture named Y–Net. The results show that the spectral bands selected by the band selection module of Y–Net achieve more reliable classification performance than traditional feature selection methods. Y–Net obtained the best classification accuracy compared to support vector machines, one–dimensional (1D) CNNs, and two–dimensional (2D) CNNs. After the network pruned the trained Y–Net, the model size was reduced to one–third of the original size, and the accuracy rate reached 98.34%. The study results can provide new ideas and references for disease identification of corn and other crops.

Novel obstacle-avoiding path planning for crop protection UAV using optimized Dubins curve

Moldy maize can produce a lot of toxins, which is harmful to human and livestock. Therefore, early detection of maize mildew is of great significance. In this study, the hyperspectral image data of maize seed with five mildew grades of the same variety were selected as the data source, by comparing a variety of preprocessing and feature extraction methods, the combination method of standard normal variate and uninformative variable elimination was selected to process hyperspectral data. In view of the shortcomings of traditional BP neural network, such as easy to fall into local optimum and slow convergence speed, BP network with ant colony optimization classification model was established by introducing ant colony optimization weight threshold. Support vector machine based on linear kernel, support vector machine based on quadratic kernel and BP neural network model were compared and the classification results were analyzed. The results show that the standard normal variate and uninformative variable elimination can effectively eliminate the error caused by solid particle surface scattering and reduce the amount of data. Among the four recognition models, BP network with ant colony optimization has the highest classification accuracy, the overall classification accuracy reaches 92.00%, which is 8.00% higher than that of the BP neural network, 12.00% higher than the support vector machine with linear kernel function and 16.00% higher than the support vector machine with quadratic kernel function, indicating that the ant colony optimization can effectively improve the recognition accuracy of the BP neural network model. This paper can provide technical support and new ideas for maize seed early mildew detection and maize seed selection.