TAPEDA

Towards Automated Post-Earthquake Damage Assessment

Started
January 9, 2019
Status
In Progress
Share this project

Abstract

After earthquakes, engineers assess the damaged state of buildings through visual inspection and judge whether they are safe to enter or not. Damage evaluation is therefore at present a very labor-intensive task, which requires a visual inspection of all potentially affected buildings. After the magnitude M=6.3 earthquake in L’Aquila in 2009, for example, 8’000 experts assessed 72’000 buildings over a period of five months spending in average 0.5 working days per building. In addition to this, the evaluation of the damage is very subjective in nature and therefore error-prone. The TAPEDA project addresses stone masonry buildings, which are among the most vulnerable buildings during earthquakes. First, we need to obtain a 3D abstraction of the building from its point cloud obtained through Structure from Motion (SFM) techniques, which helps setting up a finite element model. Second, we need detect damage to a building from images using computer vision techniques and place them on the 3D model. The capacity loss assessment may now be performed automatically using computational models. Even with some manual intervention, such a computer-assissted approach is expected to reduce the time of capacity loss from days to hours, using far fewer engineers, and be more objective in nature.

People

Collaborators

SDSC Team:
Radhakrishna Achanta

PI | Partners:

Earthquake Engineering and Structural Dynamics Laboratory:

  • Prof. Katrin Beyer

More info

description

  • The TAPEDA project is concerned with the automatic assessment of safety of buildings in the aftermath of an earthquake, which, done manually, is arduous, time-consuming, and expensive.
  • To do this automatically, we need to take pictures of a building with a drone and obtain a wireframe model of the building. From this an equivalent Frame Model (EFM) can be obtained, which permits the computation of capacity loss of the building.
  • There are two possibilities in sight to obtain a wireframe. One requires us to create the point cloud and then do meshing. The other approach may be to bypass the creation of the point cloud altogether by creating the wireframes directly from 2D pictures.

Approach

To assess safety of damaged buildings automatically, we need to take pictures of a building with a drone and obtain a wireframe abstraction of the building. We have a pipeline now which nearly gets us to the end goal but requires minor manual interventions (see figure below). This pipeline requires us to create a point cloud using Structure-from-Motion (SfM) techniques, which is reduced to a planar abstraction by fitting 3D planes to the point cloud. Alongside, damages, in particular cracks, as well as opening, namely, windows and doors are detected using deep networks and then projected on to the 3D model of the building.

Gallery

Annexe

Additionnal resources

Bibliography

Publications

Related Pages

More projects

ML4FCC

In Progress
Machine Learning for the Future Circular Collider Design
Big Science Data

CLIMIS4AVAL

In Progress
Real-time cleansing of snow and weather data for operational avalanche forecasting
Energy, Climate & Environment

SEMIRAMIS

Completed
AI-augmented architectural design
Energy, Climate & Environment

4D-Brains

In Progress
Extracting activity from large 4D whole-brain image datasets
Biomedical Data Science

News

Latest news

PassGPT | Using language models to enhance password security
February 6, 2024

PassGPT | Using language models to enhance password security

PassGPT | Using language models to enhance password security

PassGPT is a Large Language Model for password generation trained on leaked passwords, which can outperform existing methods based on generative adversarial networks by guessing twice as many unseen passwords.
ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning
January 24, 2024

ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning

ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning

Applying machine learning to ecotoxicology could help reduce the number of animal tests, costs, and animals sacrificed while preserving the accuracy of the in vivo tests.
License Flowers | Art and AI at SDSC
February 21, 2024

License Flowers | Art and AI at SDSC

License Flowers | Art and AI at SDSC

An adventure to create art using AI to raise awareness on code licenses

Contact us

Let’s talk Data Science

Do you need our services or expertise?
Contact us for your next Data Science project!