Polis: 3D Self-Supervision at City Scale

• URL: http://arxiv.org/abs/2608.29426v1
• Date: Sat, 29 Aug 2026 20:07:51 GMT
• Title: Polis: 3D Self-Supervision at City Scale
• Abstract summary: We introduce Polis, to our knowledge the first application of Sketched Isotropic Gaussian Regularization ( SIGReg) as an objective for a native point cloud encoder.<n>We evaluate it through a benchmark spanning fourteen city- and building-scale corpora.<n>Our results show that distributionally-regularized joint embedding architectures can be successful on challenging city-scale 3D scenes.
• Score: 38.40501054437541
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Reliable semantic representations derived from city-scale 3D models are increasingly important for urban analysis, infrastructure monitoring, autonomous systems, and heritage conservation. However, urban scenes of large spatial extent captured through aerial surveying differ substantially from the indoor, object-level, and self-driving LiDAR data used to pretrain most 3D self-supervised models. We introduce Polis, to our knowledge the first application of Sketched Isotropic Gaussian Regularization (SIGReg) as an objective for a native point cloud encoder, and evaluate it through a frozen-feature benchmark spanning fourteen city- and building-scale corpora. Polis combines geometrically matched cosine invariance, SIGReg, and VICReg-style anti-collapse terms with a 12.8k-scene outdoor pretraining mixture and gravity-preserving spatial view sampling. Controlled ablations show that this objective outperforms student--teacher architecture alternatives, as well as Polis versions without anti-collapse terms, on the same representative outdoor corpus. On three pretraining-disjoint city datasets, Polis reaches $23.8\%$ mean mIoU versus $16.3\%$ for the next-best encoder under high-capacity frozen probing, and $17.3\%$ versus $16.1\%$ at a matched point and voxel budget. The same city-scale lead holds on datasets whose training sets were seen in pretraining. On localized terrestrial captures with fine-grained facade and streetscape labels, the ranking reverses. Our results show that distributionally-regularized joint embedding architectures can be successful on challenging city-scale 3D scenes, and that transfer improves when self-supervision is designed for the capture geometry and spatial context of this domain while also revealing the limits of this specialization.

Related papers

• SharpSplat: Edge-Regularized 3D Gaussian Splatting for High Fidelity Urban Building Reconstruction from UAV images [8.23277995673829]
  We propose a semantic edge regularization framework that supervises 3DGS to produce crisp architectural boundaries.<n>Our method leverages SAM 3 to generate precise building masks, from which we extract architecturally significant edges.<n>Our approach proves robust across diverse building types, roof geometries, and urban densities.
  arXiv  Detail & Related papers  (2026-07-04T13:31:22Z)
• SemCityLoc: Aerial 6DoF Localization Using Semantic 3D City Models [51.121529170856064]
  We propose SemCityLoc, a semantic-geometric alignment system that reframes aerial pose estimation.<n>Instead of matching sparse contours or dense texture, our method aligns semantic surfaces and monocular depth with lightweight semantic 3D building models.<n>Experiments demonstrate substantial improvements over existing map-based approaches.
  arXiv  Detail & Related papers  (2026-06-25T18:13:39Z)
• GS4City: Hierarchical Semantic Gaussian Splatting via City-Model Priors [21.77168114492861]
  We present GS4City, a hierarchical semantic Gaussian Splatting method that incorporates city-model priors for urban scene understanding.<n>GS4City derives reliable image-aligned masks from Level of Detail 3 CityGML models via two-pass raycasting.<n>It then fuses these geometry-grounded masks with foundation-model predictions to establish scene-consistent instance correspondences.
  arXiv  Detail & Related papers  (2026-04-13T12:42:28Z)
• Unifying UAV Cross-View Geo-Localization via 3D Geometric Perception [51.687842983240564]
  Cross-view geo-localization for Unmanned Aerial Vehicles (UAVs) remains challenging due to the severe geometric discrepancy between oblique UAV imagery and satellite maps.<n>We propose a geometry-aware UAV geo-localization framework that explicitly models the 3D scene geometry to unify coarse place recognition and fine-grained pose estimation.<n>Our method significantly outperforms state-of-the-art baselines, achieving robust meter-level localization accuracy and improved generalization in complex urban environments.
  arXiv  Detail & Related papers  (2026-04-02T08:08:41Z)
• SpatialFly: Geometry-Guided Representation Alignment for UAV Vision-and-Language Navigation in Urban Environments [49.966170814478915]
  We propose a geometry-guided spatial representation framework for UAV VLN.<n>Running on RGB observations without explicit 3D reconstruction, SpatialFly introduces a geometry-guided 2D representation alignment mechanism.<n> Experimental results show that SpatialFly consistently outperforms state-of-the-art UAV VLN baselines across both seen and unseen environments.
  arXiv  Detail & Related papers  (2026-03-22T03:56:58Z)
• When the City Teaches the Car: Label-Free 3D Perception from Infrastructure [47.72611382185263]
  Building robust 3D perception for self-driving still relies heavily on large-scale data collection and manual annotation.<n>Modern cities are increasingly instrumented with roadside units (RSUs), static sensors deployed along roads and at intersections to monitor traffic.<n>We propose infrastructure-taught, label-free 3D perception, a paradigm in which RSUs act as stationary, unsupervised teachers for ego vehicles.
  arXiv  Detail & Related papers  (2026-03-17T16:21:06Z)
• Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligence [78.1406635199656]
  Holi-Spatial is the first fully automated, large-scale, spatially-aware multimodal dataset constructed from raw video inputs without human intervention.<n>Holi-Spatial-4M is the first large-scale, high-quality 3D semantic dataset, containing 12K optimized 3DGS scenes, 1.3M 2D masks, 320K 3D bounding boxes, 320K instance captions, 1.2M 3D grounding instances, and 1.2M spatial QA pairs.
  arXiv  Detail & Related papers  (2026-03-08T14:49:20Z)
• FreeOcc: Training-free Panoptic Occupancy Prediction via Foundation Models [0.9345376836714131]
  FreeOcc is a training-free pipeline that recovers semantics and geometry from multi-view images.<n>On Occ3D-nuScenes, FreeOcc achieves 16.9 mIoU and 16.5 RayIoU train-free, on par with state-of-the-art weakly supervised methods.
  arXiv  Detail & Related papers  (2026-03-06T11:19:33Z)
• GaussRender: Learning 3D Occupancy with Gaussian Rendering [86.89653628311565]
  GaussRender is a module that improves 3D occupancy learning by enforcing projective consistency.<n>Our method penalizes 3D configurations that produce inconsistent 2D projections, thereby enforcing a more coherent 3D structure.
  arXiv  Detail & Related papers  (2025-02-07T16:07:51Z)
• Semi-supervised Learning from Street-View Images and OpenStreetMap for Automatic Building Height Estimation [59.6553058160943]
  We propose a semi-supervised learning (SSL) method of automatically estimating building height from Mapillary SVI and OpenStreetMap data. The proposed method leads to a clear performance boosting in estimating building heights with a Mean Absolute Error (MAE) around 2.1 meters. The preliminary result is promising and motivates our future work in scaling up the proposed method based on low-cost VGI data.
  arXiv  Detail & Related papers  (2023-07-05T18:16:30Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.