IntroductionCrop pests have a great impact on the quality and yield of crops. The use of deep learning for the identification of crop pests is important for crop precise management.MethodsTo address the lack of data set and poor classification accuracy in current pest research, a large-scale pest data set named HQIP102 is built and the pest identification model named MADN is proposed. There are some problems with the IP102 large crop pest dataset, such as some pest categories are wrong and pest subjects are missing from the images. In this study, the IP102 data set was carefully filtered to obtain the HQIP102 data set, which contains 47,393 images of 102 pest classes on eight crops. The MADN model improves the representation capability of DenseNet in three aspects. Firstly, the Selective Kernel unit is introduced into the DenseNet model, which can adaptively adjust the size of the receptive field according to the input and capture target objects of different sizes more effectively. Secondly, in order to make the features obey a stable distribution, the Representative Batch Normalization module is used in the DenseNet model. In addition, adaptive selection of whether to activate neurons can improve the performance of the network, for which the ACON activation function is used in the DenseNet model. Finally, the MADN model is constituted by ensemble learning.ResultsExperimental results show that MADN achieved an accuracy and F1Score of 75.28% and 65.46% on the HQIP102 data set, an improvement of 5.17 percentage points and 5.20 percentage points compared to the pre-improvement DenseNet-121. Compared with ResNet-101, the accuracy and F1Score of MADN model improved by 10.48 percentage points and 10.56 percentage points, while the parameters size decreased by 35.37%. Deploying models to cloud servers with mobile application provides help in securing crop yield and quality.

The process of rapid urbanization has intensified the conversion of different land use types, resulting in a substantial loss of ecological land and ecological security being threatened. In the context of China’s vigorous advocacy of an ecological civilization, it is important to explore future land use patterns under ecological security constraints to promote sustainable development. The insufficient consideration of land ecological security in existing land use pattern simulation studies makes it difficult to effectively promote improvement in the ecological security level. Therefore, we developed a land use simulation framework that integrates land ecological security. Taking the sustainable development of land ecosystems as the core, the land ecological security index (LESI) and ecological zoning (EZ) were determined by the pressure–state–response (PSR) model and the catastrophe progression method (CPM). Natural development (ND) and ecological protection (EP) scenarios were then constructed taking the LESI and EZ into consideration. The CA–Markov model was used to simulate the land use pattern of Guangzhou for 2030 under the two scenarios. The results showed that (1) the study area was divided into four categories: ecological core zone, ecological buffer zone, ecological optimization zone, and urban development zone, with area shares of 37.53%, 31.14%, 16.96%, and 14.37%, respectively. (2) In both scenarios, the construction land around the towns showed outward expansion; compared with the ND scenario, the construction land in the EP scenario decreased by 369.10 km2, and the woodland, grassland, and farmland areas increased by 337.04, 20.80, and 10.51 km2, respectively, which significantly improved the ecological security level. (3) In the EP scenario, the construction land in the ecological core zone, ecological buffer zone, and ecological optimization zone decreased by 85.49, 114.78, and 178.81 km2, respectively, and no new construction land was added in the ecological core zone, making the land use pattern of the EP scenario more reasonable. The results of the study have confirmed that the land use pattern simulation framework integrating land ecological security can effectively predict land use patterns in different future scenarios. This study can provide suggestions and guidance for managers to use in formulating ecological protection policies and preparing territorial spatial planning.

No abstract