Machine Learning Models for Urban Image Analysis: Building Height Estimation
Description:
Artificial intelligence (AI) is triggering major advances in construction engineering in the era of the generalization of Building Information Modelling (BIM). One of the main areas of AI development is the determination of building attributes from street view images. Building height is an essential factor for the assessment of structural vulnerability that is often missing, although sometimes available in cadastral databases. The goal of this research is to help the KUK_AHPAN project (RTI2018-094827-B-C21/C22) in the collection of building attributes for seismic exposure assessment in Central American cities. The techniques presented on this paper are being applied to datasets from San José (Costa Rica). This work presents the framework and the development of machine learning techniques for urban image analysis and an application to the estimation of building heights.
Building heights were calculated from street view imagery based on a semantic segmentation machine learning model. The model has a fully convolutional architecture, and it is based on the HRNet encoder and on ResNexts depth separable convolutions, achieving fast runtime and state of the art results on standard semantic segmentation tasks. Average building heights on a pilot German street were satisfactorily estimated with a maximum error of 3 meters. Further research alternatives are commented, as well as the difficulties of obtaining valuable training data to apply these models in countries with no training datasets and different urbanism conditions. This line of research contributes to the characterization of buildings and the estimation of attributes essential for the assessment of seismic risk using automatically processed street view imagery.
Session: Opportunities and Challenges for Machine Learning Applications in Seismology [Poster]
Type: Poster
Date: 4/19/2023
Presentation Time: 08:00 AM (local time)
Presenting Author: Maria Belén Benito Oterino
Student Presenter: No
Invited Presentation:
Authors
Miguel Ureña-Pliego Corresponding Author miguel.urena@upm.es ETSI Caminos Canales y Puertos, Universidad Politécnica de Madrid |
Rubén Martínez-Marín ruben.martinez@upm.es ETSI Caminos Canales y Puertos, Universidad Politécnica de Madrid |
Beatriz González-Rodrigo beatriz.gonzalez.rodrigo@upm.es ETSI Caminos Canales y Puertos, Universidad Politécnica de Madrid |
Maria Belén Benito Oterino Presenting Author mariabelen.benito@upm.es Universidad Politécnica de Madrid |
Miguel Marchamalo-Sacristán miguel.marchamalo@upm.es ETSI Caminos Canales y Puertos, Universidad Politécnica de Madrid |
|
|
|
|
Machine Learning Models for Urban Image Analysis: Building Height Estimation
Category
Opportunities and Challenges for Machine Learning Applications in Seismology