Vectorizing World Buildings: Planar Graph Reconstruction by Primitive Detection and Relationship Inference
Abstract
This paper tackles a 2D architecture vectorization problem, whose task is to infer an outdoor building architecture as a 2D planar graph from a single RGB image. We provide a new benchmark with ground-truth annotations for 2,001 complex buildings across the cities of Atlanta, Paris, and Las Vegas. We also propose a novel algorithm utilizing 1) convolutional neural networks (CNNs) that detects geometric primitives and infers their relationships and 2) an integer programming (IP) that assembles the information into a 2D planar graph. While being a trivial task for human vision, the inference of a graph structure with an arbitrary topology is still an open problem for computer vision. Qualitative and quantitative evaluations demonstrate that our algorithm makes significant improvements over the current state-of-the-art, towards an intelligent system at the level of human perception. We will share code and data.
Cite
Text
Nauata and Furukawa. "Vectorizing World Buildings: Planar Graph Reconstruction by Primitive Detection and Relationship Inference." Proceedings of the European Conference on Computer Vision (ECCV), 2020. doi:10.1007/978-3-030-58598-3_42Markdown
[Nauata and Furukawa. "Vectorizing World Buildings: Planar Graph Reconstruction by Primitive Detection and Relationship Inference." Proceedings of the European Conference on Computer Vision (ECCV), 2020.](https://mlanthology.org/eccv/2020/nauata2020eccv-vectorizing/) doi:10.1007/978-3-030-58598-3_42BibTeX
@inproceedings{nauata2020eccv-vectorizing,
title = {{Vectorizing World Buildings: Planar Graph Reconstruction by Primitive Detection and Relationship Inference}},
author = {Nauata, Nelson and Furukawa, Yasutaka},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
year = {2020},
doi = {10.1007/978-3-030-58598-3_42},
url = {https://mlanthology.org/eccv/2020/nauata2020eccv-vectorizing/}
}