BRAILS (Building and Infrastructure Recognition using AI at Large-Scale) provides a set of Python modules that utilize deep learning (DL), and computer vision (CV) techniques to extract information from satellite and street level images. The BRAILS framework also provides turn-key applications allowing users to put individual modules together to determine multiple attributes in a single pass or train general-purpose image classification, object detection, or semantic segmentation models.
Online documentation is available at https://nheri-simcenter.github.io/BRAILS-Documentation.
The easiest way to install the latest version of BRAILS is using pip
:
pip install BRAILS
This example demonstrates how to use the InventoryGenerator
method embedded in BRAILS to generate regional-level inventories.
The primary input to InventoryGenerator
is location. InventoryGenerator
accepts four different location
input types: 1) region name, 2) list of region names, 3) a tuple containing the coordinates for two opposite vertices of a bounding box for a region (e.g., (vert1lon,vert1lat,vert2lon,vert2lat)
), and a 4) GeoJSON file containing building footprints or location points.
InventoryGenerator automatically detects building locations in a region by downloading footprint data for the location
input. The three footprint data sources, fpSource
, included in BRAILS are i) OpenStreetMaps, ii) Microsoft Global Building Footprints dataset, and iii) FEMA USA Structures. The keywords for these sources are osm
, ms
, and usastr
, respectively.
InventoryGenerator
also allows inventory data to be imported from the National Structure Inventory or another user-specified file to create a baseline building inventory.
Please note that to run the generate
method of InventoryGenerator
, you will need a Google API Key.
#import InventoryGenerator:
from brails.InventoryGenerator import InventoryGenerator
# Initialize InventoryGenerator:
invGenerator = InventoryGenerator(location='Berkeley, CA',
fpSource='usastr',
baselineInv='nsi',
lengthUnit='m',
outputFile='BaselineInvBerkeleyCA.geojson')
# View a list of building attributes that can be obtained using BRAILS:
invGenerator.enabled_attributes()
# Run InventoryGenerator to generate an inventory for the entered location:
# To run InventoryGenerator for all enabled attributes set attributes='all':
invGenerator.generate(attributes=['numstories','roofshape','buildingheight'],
GoogleAPIKey='ENTER-YOUR-API-KEY-HERE',
nbldgs=100,
outputFile='BldgInvBerkeleyCA.geojson')
# View generated inventory:
invGenerator.inventory
This work is based on material supported by the National Science Foundation under grants CMMI 1612843 and CMMI 2131111.
NHERI-SimCenter nheri-simcenter@berkeley.edu
@software{cetiner_2024_10448047,
author = {Barbaros Cetiner and
Charles Wang and
Frank McKenna and
Sascha Hornauer and
Jinyan Zhao and
Yunhui Guo and
Stella X. Yu and
Ertugrul Taciroglu and
Kincho H. Law},
title = {BRAILS Release v3.1.1},
month = feb,
year = 2024,
publisher = {Zenodo},
version = {v3.1.1},
doi = {10.5281/zenodo.10606032},
url = {https://doi.org/10.5281/zenodo.10606032}
}