This document explains how to use the Pedestrian Infrastructure Classification pretrained model available on ArcGIS Living Atlas of the World. The model is used to classify pixels of crosswalks, sidewalks/footpaths, and roads from high-resolution satellite RGB imagery.
Pedestrian infrastructure includes sidewalks, crosswalks, and other features that facilitate safe and accessible walking routes. As urban areas continue to expand, the quality and availability of such infrastructure directly impact pedestrian safety, mobility, and overall urban livability. The Pedestrian Infrastructure Classification model can be used to create detailed maps of sidewalks, crosswalks, and road networks that can aid in urban planning. Such data can also help in smart city initiatives, disaster response, and sustainable and economic development by enabling the creation of more walkable and inclusive urban environments.
The classifications from the model can be used to create pedestrian networks that can support applications such as pedestrian routing, flow analysis, and the planning of step-free access and vision zero initiatives. The model can also be used by planners to enhance urban safety by identifying hazardous areas and improve accessibility for vulnerable users such as the elderly and disabled. The model can provide precise data that can assist in optimizing public transportation and reduce CO2 emissions. As cities prioritize pedestrian-friendly environments to address climate change, public health, and economic competitiveness, this model provides a scalable, cost-effective solution to drive sustainable urban development.
License requirements
To complete this workflow, the following are the license requirements:
- ArcGIS Desktop—ArcGIS Image Analyst extension for ArcGIS Pro
- ArcGIS Enterprise—ArcGIS Image Server with raster analytics configured
- ArcGIS Online—ArcGIS Pro or Professional Plus user type
Model overview
This model has the following characteristics:
- Input—8-bit, RGB high-resolution (10 - 30 centimeters) imagery.
- Output—Classified raster layer representing classified roads, crosswalks, and sidewalks. Everything else is classified as background class.
- Compute—This workflow is compute-intensive, and a GPU with a minimum CUDA compute capability of 6.0 is recommended.
- Applicable geographies—This model is expected to work well in the following states in the U.S., as described here:
State Entire state CA
False
DC
True
MA
True
NY
True
NJ
True
OR
True
TN
False
VA
True
WA
False
- Architecture—This model is based on the Tile2Net model architecture.
- Accuracy metrics—The model
has
the following
metrics on the
test data
from
Boston,
Cambridge, Washington
D.C.,
and
Manhattan:
Label IoU (%) Precision Recall Sidewalk
82.67
0.90
0.92
Road
86.04
0.91
0.94
Crosswalk
75.42
0.86
0.86
Background
93.94
0.97
0.96
mIoU (%)
84.51
- Reference—
Hosseini, M., Sevtsuk, A., Miranda, F., Cesar, R. M., & Silva, C. T. (2023). Mapping the walk: A scalable computer vision approach for generating sidewalk network datasets from aerial imagery. Computers Environment and Urban Systems, 101, 101950. https://doi.org/10.1016/j.compenvurbsys.2023.101950
Access and download the model
Download the Pedestrian Infrastructure Classification pretrained model from ArcGIS Living Atlas of the World. Alternatively, access the model directly from ArcGIS Pro, or consume it in ArcGIS Image for ArcGIS Online.
- Browse to ArcGIS Living Atlas of the World.
- Sign in with your ArcGIS Online credentials.
- Search for Pedestrian Infrastructure Classification and open the item page from the search results.
- Click the Download button to download the model.
You can use the downloaded .dlpk file directly in ArcGIS Pro or upload and use it in ArcGIS Enterprise.
Release notes
The following are the release notes:
| Date | Description |
|---|---|
| December 2024 | First release of Pedestrian Infrastructure Classification |