Building segmentation from remote sensing data has been boosted in recent years by the advances in data processing algorithms and the improved ability of sensors to capture very high spatial resolution (VHSR) images. The prevailing approaches to this task are built upon deep semantic segmentation networks learned on ortho geometry, i.e., orthorectified images. In this study, we propose to deal with the building segmentation using multiview Pléiades satellite images at the perfect sensor (PS) geometry instead of the ortho geometry. This has two main advantages: 1) it frees the segmentation workflow from geometric imprecision that may arise after the image rectification step and 2) it allows the physical augmentations of the ground truth (GT) by reprojecting building objects on each of the multiview acquisitions. The GT reprojection process makes use of rational polynomial coefficients (RPCs) provided as image metadata and a fine scale digital surface model (DSM). We assess the benefit of our proposal using a U-Net encoder-decoder learned on a dataset composed of tri-stereo Pléiades acquisitions over six French cities. Experimental results demonstrate the significance of the proposal especially an enhanced generalization capability for the building segmentation.

We discuss the applicability of large quantum annealers for the purpose of processing Remote Sensing images. We show an application of currently existing quantum annealers for the purpose of post-processing segmentation of a hyperspectral image. We show that in principle there might exist useful applications of large scale quantum annealers for the purpose of Remote Sensing data processing.

Supervised machine learning techniques are widely used for hyper-spectral images segmentation. A typical simple scheme of classification of such images probabilistically assigns a label to each individual pixel omitting information about pixel surroundings. In order to achieve better classification results for real world images one has to agree the local label obtained from the classifier with the classes of pixel neighborhood. A popular way to do it is through a probabilistic graphical model, where label distributions for individual pixels are mapped into a graph of neighborhood relations. One way to realize this approach is to use Ising models, where class probability is mapped to spin energy and class-class interaction is mapped to the spins coupling. By finding low energy states of such an Ising model we can perform post-processing of segmented images. In this work we present how this postprocessing can be implemented using a quantum annealer.

This article describes a practical Earth Observation use case that would benefit from quantum computing. We analyze three quantum neural network algorithms. We implemented one of the algorithms on the EuroSAT dataset. We compare the algorithms with respect to complexity and degree of quantization. We believe that the algorithms we propose would be useful for the remote sensing community when quantum computing technologies become widely available.1

3D Geospatial information plays a key role in many soaring sectors such as sustainable and smart cities, climate monitoring, ecological mobility, and economic intelligence. The availability of huge volumes of satellite, airborne and insitu data now makes this production feasible at large scale. It needs nonetheless a certain level of manual intervention to secure the level of quality, which prevents mass production. This paper presents the AI4GEO program that aims at developing an end to end solution to produce automatically qualified 3D Digital model at scale together with multiple layers of information.