Research on Intelligent Recognition of Remote Sensing Images Based on U-Net Neural Network Algorithm

At present, there are many application scenarios for remote sensing image recognition in water conservancy, hydropower, energy and other engineering industries. The recognition algorithm under the existing technology usually has the problems of low efficiency of image data sample production and low efficiency and accuracy of image recognition. In this paper, an algorithm model based on semantic segmentation neural network is studied. The image data augmentation network module and block weight voting mechanism are introduced on the basis of U-Net network, which reduces the risk of overfitting in the training of the algorithm model and improves the accuracy and robustness of the model. At the same time, the interpretability and accuracy of the prediction are significantly improved by the method of image segmentation and weight voting. After the training, the algorithm is tested and applied to real remote sensing images. The experimental results show that the algorithm can improve the recognition efficiency and accuracy of remote sensing image to a great extent, and has a great application prospect in the engineering industry involving remote sensing image.