Sci-Net: Scale Invariant Model for Buildings Segmentation from Aerial Imagery
This addresses a practical issue for users in earth observation who deal with diverse image resolutions, offering a scale-invariant solution, though it appears incremental as it builds on existing architectures like UNet.
The paper tackles the problem of building segmentation from aerial imagery across varying spatial resolutions, proposing Sci-Net to overcome performance degradation from re-sampling, and reports that it significantly outperforms state-of-the-art models on datasets like Open Cities AI and Multi-Scale Building with steady improvements.
Buildings' segmentation is a fundamental task in the field of earth observation and aerial imagery analysis. Most existing deep learning-based methods in the literature can be applied to a fixed or narrow-range spatial resolution imagery. In practical scenarios, users deal with a broad spectrum of image resolutions. Thus, a given aerial image often needs to be re-sampled to match the spatial resolution of the dataset used to train the deep learning model, which results in a degradation in segmentation performance. To overcome this challenge, we propose, in this manuscript, Scale-invariant Neural Network (Sci-Net) architecture that segments buildings from wide-range spatial resolution aerial images. Specifically, our approach leverages UNet hierarchical representation and Dense Atrous Spatial Pyramid Pooling to extract fine-grained multi-scale representations. Sci-Net significantly outperforms state of the art models on the Open Cities AI and the Multi-Scale Building datasets with a steady improvement margin across different spatial resolutions.