 This paper proposes a novel pipeline for building segmentation from Sentinel-2 data with 10M spatial resolution. It combines a super-resolution, SR, component with a semantic segmentation component. The SR component improves the quality of the infrastructure analysis through medium-resolution satellite data. Additionally, a unique dataset for the Russian Federation covering an area of 1,091.2 square kilometers was collected and made available. The dataset includes Sentinel-2 imagery adjusted to the spatial resolution of 2.5M and is accompanied by semantic segmentation masks. Experiments were conducted for the SR task, using advanced image SR methods such as the diffusion-based SR-3 model, RCN, SR-GEN, and MCGR. The MCGR network produced the best result, with a PSNR of 27.54 and SSM of 0.79. The obtained SR images were then used to tackle the building segmentation task with different neural network models, including D-Lab V3 with different encoders, SWIN, and TWINZ transformers. The SWIN transformer achieved the best results, with an F1 score of 79. This article was authored by Svetlana Alaryanova, Dmitry Shadrin, Islan Jon Shukratov, and others. We are article.tv, links in the description below.