-
Polygonizer: An auto-regressive building delineator
Authors:
Maxim Khomiakov,
Michael Riis Andersen,
Jes Frellsen
Abstract:
In geospatial planning, it is often essential to represent objects in a vectorized format, as this format easily translates to downstream tasks such as web development, graphics, or design. While these problems are frequently addressed using semantic segmentation, which requires additional post-processing to vectorize objects in a non-trivial way, we present an Image-to-Sequence model that allows…
▽ More
In geospatial planning, it is often essential to represent objects in a vectorized format, as this format easily translates to downstream tasks such as web development, graphics, or design. While these problems are frequently addressed using semantic segmentation, which requires additional post-processing to vectorize objects in a non-trivial way, we present an Image-to-Sequence model that allows for direct shape inference and is ready for vector-based workflows out of the box. We demonstrate the model's performance in various ways, including perturbations to the image input that correspond to variations or artifacts commonly encountered in remote sensing applications. Our model outperforms prior works when using ground truth bounding boxes (one object per image), achieving the lowest maximum tangent angle error.
△ Less
Submitted 8 April, 2023;
originally announced April 2023.
-
Learning to Generate 3D Representations of Building Roofs Using Single-View Aerial Imagery
Authors:
Maxim Khomiakov,
Alejandro Valverde Mahou,
Alba Reinders Sánchez,
Jes Frellsen,
Michael Riis Andersen
Abstract:
We present a novel pipeline for learning the conditional distribution of a building roof mesh given pixels from an aerial image, under the assumption that roof geometry follows a set of regular patterns. Unlike alternative methods that require multiple images of the same object, our approach enables estimating 3D roof meshes using only a single image for predictions. The approach employs the PolyG…
▽ More
We present a novel pipeline for learning the conditional distribution of a building roof mesh given pixels from an aerial image, under the assumption that roof geometry follows a set of regular patterns. Unlike alternative methods that require multiple images of the same object, our approach enables estimating 3D roof meshes using only a single image for predictions. The approach employs the PolyGen, a deep generative transformer architecture for 3D meshes. We apply this model in a new domain and investigate the sensitivity of the image resolution. We propose a novel metric to evaluate the performance of the inferred meshes, and our results show that the model is robust even at lower resolutions, while qualitatively producing realistic representations for out-of-distribution samples.
△ Less
Submitted 20 March, 2023;
originally announced March 2023.
-
SolarDK: A high-resolution urban solar panel image classification and localization dataset
Authors:
Maxim Khomiakov,
Julius Holbech Radzikowski,
Carl Anton Schmidt,
Mathias Bonde Sørensen,
Mads Andersen,
Michael Riis Andersen,
Jes Frellsen
Abstract:
The body of research on classification of solar panel arrays from aerial imagery is increasing, yet there are still not many public benchmark datasets. This paper introduces two novel benchmark datasets for classifying and localizing solar panel arrays in Denmark: A human annotated dataset for classification and segmentation, as well as a classification dataset acquired using self-reported data fr…
▽ More
The body of research on classification of solar panel arrays from aerial imagery is increasing, yet there are still not many public benchmark datasets. This paper introduces two novel benchmark datasets for classifying and localizing solar panel arrays in Denmark: A human annotated dataset for classification and segmentation, as well as a classification dataset acquired using self-reported data from the Danish national building registry. We explore the performance of prior works on the new benchmark dataset, and present results after fine-tuning models using a similar approach as recent works. Furthermore, we train models of newer architectures and provide benchmark baselines to our datasets in several scenarios. We believe the release of these datasets may improve future research in both local and global geospatial domains for identifying and map** of solar panel arrays from aerial imagery. The data is accessible at https://osf.io/aj539/.
△ Less
Submitted 2 December, 2022;
originally announced December 2022.
