Texture2LoD3: Enabling LoD3 Building Reconstruction With Panoramic Images

CVPRW 2025, Workshop on Urban Scene Modeling

Wenzhao Tang*1, Weihang Li*1,2, Xiucheng Liang3, Olaf Wysocki1, Filip Biljecki3, Christoph Holst1, Boris Jutzi1
1Technical University of Munich    2Munich Center for Machine Learning    3National University of Singapore   
*Wenzhao Tang and Weihang Li contributed equally to this work.
{wenzhao.tang, weihang.li, olaf.wysocki}@tum.de
Paper arXiv Code Datasets
Input Image 1

Texture2LoD3 proposes leveraging ubiquitous street-level images and low-level building models for accurate ortho-texturing (left): Enabling accurate semantic segmentation (center) and facade-rich LoD3 reconstruction (right).

Abstract

Despite recent advancements in surface reconstruction, Level of Detail (LoD) 3 building reconstruction remains an unresolved challenge. The main issue pertains to the object-oriented modelling paradigm, which requires georeferencing, watertight geometry, facade semantics, low-poly representation, and starkly contrasting unstructured mesh-oriented models. In Texture2LoD3, we introduce a novel method leveraging the ubiquity of 3D building model priors and panoramic street-level images, enabling the reconstruction of LoD3 building models. We observe that prior low-detail building models can serve as valid planar targets for ortho-rectifying street-level panoramic images. Moreover, deploying segmentation on rectified images on low-level building surfaces supports maintaining essential georeferencing, watertight geometry, and low-poly representation for LoD3 reconstruction. In the absence of LoD3 validation data, we additionally introduce the ReLoD3 dataset, on which we experimentally demonstrate that our method leads to improved facade segmentation accuracy by 11% and can replace costly manual projections. We believe that Texture2LoD3 can scale the adoption of LoD3 models, opening applications in estimating building solar potential or enhancing autonomous driving simulations.

Method

teaser-fig.

Overview of the proposed Texture2LoD3 method: The method commences with global matching of georeferenced panorama images and low-level 3D models. In the top branch, 3D target facade surfaces are simplified, while in the bottom branch panoramic images are rectified and building facade instances are extracted. Subsequently, fine object-to-object matching and projection is performed to the simplified 3D model surface. Quadrilateral fitting and image-to-plane ray casting ensure accurate ortho-rectified 3D texture, enabling accurate facade elements segmentation and LoD3 reconstruction.

ReLoD3 Texture Dataset Benchmark on Facade Segmentation

teaser-fig.

Tested facade segmentation baselines on a selected building from the introduced ReLoD3 benchmark dataset across various texture projection methods. Our Texture2LoD3 is less prone to distortions, hence yielding more accurate segmentation across the baselines.

teaser-fig.

Texture2LoD3 maintains on-par accuracy with manual texturing even in the presence of increasing facade width, unlike the method without rectification. Shown on MaskFormer on four width-different facades of the introduced ReLoD3 benchmark dataset.

BibTeX


      @article{tang2025texture2lod3,
        title={Texture2LoD3: Enabling LoD3 Building Reconstruction With Panoramic Images},
        author={Tang, Wenzhao and Li, Weihang and Liang, Xiucheng and Wysocki, Olaf and Biljecki, Filip and Holst, Christoph and Jutzi, Boris},
        journal={arXiv preprint arXiv:2504.05249},
        year={2025}
      }