Building Extraction from High Resolution Remote Sensing Images Based on Improved Mask R-CNN

In order to improve the accuracy of building extraction, an improved Mask R-CNN instance segmentation model based on deep learning is proposed in this paper. The convolutional block attention module (CBAM) is introduced into the feature extraction network to improve the significance of the features of the building area in the feature map. Augmented FPN (AugFPN) is used to replace FPN network, so as to reduce the lack of information in the process of feature fusion and further tap the potential of multi-scale features, so as to improve the accuracy of building extraction. Using this model, the improved algorithm is compared with the original algorithm in the marked dataset. The detection AP50 of the improved algorithm for buildings is 80.1%, and the segmentation AP50 is 75.9%, which are better than the original algorithm. The experimental results show that this algorithm improves the learning ability of the model on building features, and this method can effectively improve the accuracy of building extraction from high resolution remote sensing images.