https://github.com/robinlovelace/roadwidths

Compute widths of roads from OSM data
https://github.com/robinlovelace/roadwidths

Last synced: about 1 month ago
JSON representation

Compute widths of roads from OSM data

• Host: GitHub
• URL: https://github.com/robinlovelace/roadwidths
• Owner: Robinlovelace
• Created: 2022-12-29T23:42:09.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-12-30T10:44:45.000Z (over 2 years ago)
• Last Synced: 2025-04-03T16:12:21.694Z (about 2 months ago)
• Size: 4.88 KB
• Stars: 4
• Watchers: 4
• Forks: 0
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        


The aim of this repo is to document the process of estimating road widths from aerial photography.

That is a worthy aim, as evidenced by feedback to a Tweet asking how to do it: https://twitter.com/robinlovelace/status/1608595571635556352 and this toot: https://fosstodon.org/@robinlovelace/109598027567544977

The approach proposed is as follows:

- Get high resolution imagery, e.g. 25 cm resolution imagery from the Environment Agency via Edina, or open data from Switzerland in GEE, shown below.

- Create a training dataset by classifying pixels for a sample dataset, e.g. into carriageway, sidewalk (pavement), other. This can be done manually or with reference to other datasets such as MM Topo.

- Download OSM highway data for the area

- For each segment sample at intervals along the segment to get width estimates per OSM way

- Compare the estimated widths with a ground truth dataset (e.g. OSM MM Highways) to generate confidence bands for predictions (e.g. 95% CIs)

![](https://user-images.githubusercontent.com/1825120/210060557-16d93c23-c840-41f5-9d3b-2da4c4eb3d97.png)

## References

Much of the groundwork has already been done.

- The most promising repo for image classification seems to be https://github.com/jdalrym2/road_surface_classifier

- There is also https://github.com/jiankang1991/road_extraction_remote_sensing

- Tweet: https://twitter.com/andymaclachlan/status/1608574374008991745

- Tutorial on image segmentation: https://www.tensorflow.org/tutorials/images/segmentation