MonteFloor: Extending MCTS for Reconstructing Accurate Large-Scale Floor Plans

Stekovic, Sinisa; Rad, Mahdi; Fraundorfer, Friedrich; Lepetit, Vincent

doi:10.48550/arxiv.2103.11161

Search citation statements

Order By: Relevance

Paper Sections

Select...

Losses1

Citation Types

Supporting

Mentioning

Contrasting

Year Published

2021

Publication Types

Select...

Other1

Relationship

Self Cite1

Independent0

Authors

Journals

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We generate the footprint mask of the predicted polygons exploiting a Differentiable Polygon Rendering method [33]. It is the soft version of the winding number algorithm, that checks whether a pixel location x is inside the polygon P k with the equation:…”

Section: Lossesmentioning

confidence: 99%

PolyWorld: Polygonal Building Extraction with Graph Neural Networks in Satellite Images

Zorzi¹,

Bazrafkan²,

Habenschuss³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Most state-of-the-art instance segmentation methods produce binary segmentation masks, however, geographic and cartographic applications typically require precise vector polygons of extracted objects instead of rasterized output. This paper introduces PolyWorld, a neural network that directly extracts building vertices from an image and connects them correctly to create precise polygons. The model predicts the connection strength between each pair of vertices using a graph neural network and estimates the assignments by solving a differentiable optimal transport problem. Moreover, the vertex positions are optimized by minimizing a combined segmentation and polygonal angle difference loss. PolyWorld significantly outperforms the state-of-the-art in building polygonization and achieves not only notable quantitative results, but also produces visually pleasing building polygons. Code and trained weights will be soon available on github.

show abstract

Section: Lossesmentioning

confidence: 99%