In the field of remote sensing, the classification of land cover is a pivotal and challenging issue. Standard models fail to capture global and semantic information in remote sensing images despite the fact that a convolutional neural network provides robust support for semantic segmentation. In addition, owing to disparities in semantic levels and spatial resolution, the simple fusion of low-level and high-level features may diminish the efficiency. To address these deficiencies, an attention-guided multi-level feature fusion network (AMFFNet) is proposed in this study. The proposed AMFFNet approach is designed as an encoder-decoder network with the inclusion of a multi-level feature fusion module (MFF) and a dual attention map module (DAM). A DAM models the semantic association of features from a spatial and channel perspective, and an MFF bridges the semantic and resolution gaps between high-level and low-level features. Furthermore, we propose a residual-based boundary refinement upsample module to further optimize the object boundaries. The experimental results indicate that the proposed strategy can considerably enhance the accuracy of land cover classification, achieving a mean intersection over union of 90.39% on the LandCover.ai dataset and 63.14% on the Gaofen Image Dataset with 15 categories (GID-15).

This paper takes the automatic classification of the large-scale Uyghur text collected from the network as research background, designed the functional block structure of the Uyghur text classification system, and chose the KNN algorithm as the classification engine, and programmed the classification system using C sharp. In the preprocessing part, combining with the Uyghur language's lexical characteristics, we introduced the stem extraction method into the procedure, and then have greatly reduced the whole feature dimensions. the classification experimental results on the basis of large-scale text corpus includes more than 3000 documents which are belongs to different 10 categories are given, and the results of the classification experiments for the different number of features selected by using x2 statistical method are also given. The results show that only 3% to 5% of the whole high dimensional features are crucial to higher classification accuracy, so it is possible how to determine what those best features are or further reducing the feature space dimensions which are the interesting issues to be further continued.

No abstract

No abstract