Large Scale Asset Extraction for Urban Images

Affara, Lama; Nan, Liangliang; Ghanem, Bernard; Wonka, Peter

doi:10.1007/978-3-319-46487-9_27

Cited by 7 publications

(12 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In comparison, SEGNET acheived precision of 36% and recall of 28%, with an F 1 of 0.32. The recent method of Affara et al [2016] had a per-object precision of 85%, recall of 52%, and an F 1 score of 0.64 on the same data.…”

Section: Training and Evaluationmentioning

confidence: 86%

“…This network was trained on urban street scenes using CamVid data [Brostow et al 2008] and then refined using CityScapes data [Cordts et al 2016] to identify parts of images that are likely to have façade features. We then rectify based on the edges within that region using the method proposed by Affara et al [2016].…”

Section: Analyzing Street-level Imagerymentioning

confidence: 99%

“…We refine the probabilistic Bayesian SEGNET architecture to segment a set of labels for architectural façade element features using the CMP Facade dataset [Tylecek 2012], the dataset used by Affara et al [2016], and an additional dataset of 800 facades that we annotated directly from GSV images of London, Oviedo, and New York. We use this SEGNET-FACADE model (available at: https://github.com/jfemiani/facade-segmentation) to assign perpixel probabilities to the images for each feature class.…”

Section: Analyzing Street-level Imagerymentioning

confidence: 99%

See 2 more Smart Citations

BigSUR

et al. 2017

Self Cite

View full text Add to dashboard Cite

Fig. 1. Structured Urban Reconstruction. Given street-level imagery, GIS footprints, and a coarse 3D mesh (left), we formulate a global optimization to automatically fuse these noisy, incomplete, and conflicting data sources to create building footprints (middle: colored horizontal polygons) with profiles (vertical ribbons shown for several footprints) and attached building façades (vertical rectangles). The output encodes a structured urban model (right) including the walls, roof, and associated building elements (e.g., windows, balconies, roof, wall color, etc.). Inset below: A reference aerial image.The creation of high-quality semantically parsed 3D models for dense metropolitan areas is a fundamental urban modeling problem. Although recent advances in acquisition techniques and processing algorithms have resulted in large-scale imagery or 3D polygonal reconstructions, such data-sources are typically noisy, and incomplete, with no semantic structure. In this paper, we present an automatic data fusion technique that produces high-quality structured models of city blocks. From coarse polygonal meshes, street-level imagery, and GIS footprints, we formulate a binary integer program that globally balances sources of error to produce semantically parsed mass models with associated façade elements. We demonstrate our system on four city regions of varying complexity; our examples typically contain densely built urban blocks spanning hundreds of buildings. In our largest example, we produce a structured model of 37 city blocks spanning a total of 1,011 buildings at a scale and quality previously impossible to achieve automatically.

show abstract

Section: Training and Evaluationmentioning

confidence: 86%

Section: Analyzing Street-level Imagerymentioning

confidence: 99%

Section: Analyzing Street-level Imagerymentioning

confidence: 99%

See 1 more Smart Citation

BigSUR

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…Multiple low level classifiers were evaluated in the ATLAS framework [11], but modern deep learning methods were not included. Alternately, it is possible to extract boxes of labeled regions using object detection algorithms [11,17,18].…”

Section: Traditional Fac ¸Ade Parsing Methodsmentioning

confidence: 99%

Facade Segmentation in the Wild

Femiani¹,

Para²,

Mitra³

et al. 2018

Preprint

Self Cite

View full text Add to dashboard Cite

Urban fac ¸ade segmentation from automatically acquired imagery, in contrast to traditional image segmentation, poses several unique challenges. 360 • photospheres captured from vehicles are an effective way to capture a large number of images, but this data presents difficult-to-model warping and stitching artifacts. In addition, each pixel can belong to multiple fac ¸ade elements, and different facade elements (e.g., window, balcony, sill, etc.) are correlated and vary wildly in their characteristics. In this paper, we propose three network architectures of varying complexity to achieve multilabel semantic segmentation of fac ¸ade images while exploiting their unique characteristics. Specifically, we propose a MULTIFACSEGNET architecture to assign multiple labels to each pixel, a SEPARABLE architecture as a low-rank formulation that encourages extraction of rectangular elements, and a COMPATIBILITY network that simultaneously seeks segmentation across facade element types allowing the network to 'see' intermediate output probabilities of the various fac ¸ade element classes. Our results on benchmark datasets show significant improvements over existing fac ¸ade segmentation approaches for the typical fac ¸ade elements. For example, on one commonly used dataset the accuracy scores for window (the most important architectural element) increases from 0.91 to 0.97 percent compared to the best competing method, and comparable improvements on other element types.

show abstract

“…Affara et al [35] rectifies fac ¸ade images using RANSAC to estimate a homography that maps a large number of edges to vertical or horizontal lines. Wu [36] uses symmetry of repeated structures on a fac ¸ade (in addition to VPs) to rectify fac ¸ade images.…”

Section: Pose Extraction and Rectificationmentioning

confidence: 99%

Large-Scale Architectural Asset Extraction from Panoramic Imagery

Zhu

Para

Fruehstueck

et al. 2022

IEEE Trans. Visual. Comput. Graphics

Self Cite

View full text Add to dashboard Cite

We present a system to extract architectural assets from large-scale collections of panoramic imagery. We automatically rectify and crop parts of the panoramic image that contain dominant planes, and then use object detection to extract assets such as fac ¸ades and windows. We also provide various tools to identify attributes of the assets to determine the asset quality and index the assets for search. In addition, we propose a UI to visualize and query assets. Finally, we present applications for urban modeling and texture synthesis.

show abstract

Large Scale Asset Extraction for Urban Images

Cited by 7 publications

References 27 publications

BigSUR

BigSUR

Facade Segmentation in the Wild

Large-Scale Architectural Asset Extraction from Panoramic Imagery

Contact Info

Product

Resources

About