Introducing artificial intelligence technology to plant disease management for sustainable agriculture

Plant disease management plays a critical role in sustainable agricultural development. By effectively managing and controlling plant diseases, crop losses can be reduced and the quality and quantity of agricultural products can be improved. Artificial intelligence technology can quickly and accurately identify plant diseases. In this paper, we apply artificial intelligence technology to recognize tea leaf disease categories and realize the management of tea leaf diseases to promote the sustainable development of agricultural production. First, we constructed a small tea leaf disease dataset for the training and testing of artificial neural network. And then, we used MobileNet to optimize the UNet, and constructed Lightweight UNet Optimized with MobileNet (LW-MUNet) for segmenting tea leaf in pictures to improve the recognition accuracy. Finally, we proposed Multiscale Group Attention Embedded Residual Networks (MA-ResNet) to classify the diseases in segmented tea leaf images. Through experiments MA-ResNet obtains 97.43% accuracy with a lightweight structure and outperforms multiple popular models.