Examples
Explore a series of introductory examples to learn how to use BRAILS++ effectively. These examples are organized progressively, building on concepts introduced in previous examples to deepen understanding step-by-step. Each example presents one or more Python scripts, explaining their inputs and discussing the outputs generated by running the scripts. While the examples focus on key outputs, they represent only a subset of the information produced. Since BRAILS++ primarily focuses on creating regional inventories, several Jupyter notebooks are included to display interactive maps, providing insights into the larger inventories generated.
Introductory Examples
Learn how to dynamically import BRAILS++ modules for use in your pipeline.
Use BRAILS++ to extract building footprint data across broad geographic areas.
Retrieve baseline building attributes from the National Structure Inventory (NSI) using BRAILS++.
Acquire satellite and street-level imagery for buildings using BRAILS++’s automated tools.
Automatically filter out images with poor quality or limited visual detail using BRAILS++.
Fill in missing building information by applying statistical imputation methods available in BRAILS++.
Examples Gallery
The examples above give you a glimpse of how BRAILS++ can generate inventories for the built environment. But that is just the beginning, BRAILS++ has many more ways to capture and work with data. Below, you will find a gallery of hands-on examples that let you explore these capabilities in action.
Create Regional-Scale Inventories
Generate comprehensive building inventories by fusing multiple data sources and imagery to support seismic damage assessment and loss analysis.
Create detailed building inventories by combining multiple data sources and imagery, enabling accurate hurricane damage assessment and loss analysis.
Generate comprehensive inventories of power networks.
Data Scraping
Learn how to quickly fetch footprint data for any region by simply providing its name.
Learn how to extract footprint data for a region by specifying the coordinates of its bounding polygon.
Use BRAILS++ to easily access and download both aerial and street-level imagery for any region of interest. These datasets can be combined with computer vision models to extract attributes that are otherwise not available in standard databases.
Retrieve baseline building data from the NSI database to support inventory creation.
Access USGS 3DEP elevation data.
Prepare RAPID datasets for seamless integration with BRAILS++’s computer vision models, enabling automated analysis of building and infrastructure imagery.
Attribute Extraction from Imagery
Classify building construction types using street-level imagery.
Use CLIP to identify the construction type of buildings from street-level imagery.
Analyze building facades to automatically predict window areas, building height, roof height, and roof pitch angle from street-level imagery.
Automatically detect the presence of garages in buildings using street-level imagery.
Easily spot chimneys in buildings using street-level imagery.
Identify the type of building foundation from street-level images to support damage and loss predictions.
Automatically detect the number of floors in a building from street-level images using a custom object detection model.
Use GPT-powered models to predict the number of floors in buildings from street-level images.
Leverage CLIP (VLM) to classify the number of floors in buildings from street-level for automated regional studies.
Predict building occupancy types from street-level images to understand how individual spaces are used in a region.
Quickly identify building roof shapes from aerial imagery using a custom image classification model.
Use a GPT-powered model to classify roof shapes from aerial images for urban analysis and design studies.
Leverage CLIP (VLM) to classify roof shapes in aerial imagery of buildings for large-scale rooftop analysis.
Segment building images into meaningful regions using SAM for detailed attribute extraction.
Estimate the construction year of buildings from street-level images to provide temporal context in datasets.
Completing Datasets with Imputation and Ruleset-Based Logic
Fill in missing data automatically using statistical imputation techniques, ensuring your datasets are complete and ready for damage and loss analysis.
Automatically predict the building attributes needed for HAZUS flood analysis from a known set of attributes to prepare complete datasets for damage and loss assessments. These rulesets were developed by Prof. Kijewski-Correa and her team at the University of Notre Dame.
Easily predict the building attributes needed for HAZUS wind analysis using a known set of input data to quickly prepare complete datasets for damage and loss assessment studies. These rulesets were developed by Prof. Kijewski-Correa and her team at the University of Notre Dame.