OpenStreetMap
How SafeStreets uses OSM to score pedestrian safety, and what's missing in Southeast Asia
Posted by sarath sabarish on 3 March 2026 in English.Nimman Road, Chiang Mai(Thailand) is a well-mapped, high-traffic corridor. It scores a B on network density: good intersection frequency, reasonable block lengths. But it scores near zero on crossing coverage because there are no highway=crossing nodes tagged within the 800m analysis radius. The street has physical crossings. They’re just invisible to any tool that relies on OSM, which is most tools.
That’s what SafeStreets shows: not just a score, but which data gap is causing it.
What SafeStreets is?
A free tool that scores the walkability and pedestrian safety of any street address globally(graded out of 10). No account required, 190+ countries. OSM is the backbone, and the only data source that works everywhere.
How OSM powers it, three functions?
- Address geocoding via Nominatim Every analysis starts here, with a ~50km geolocation bias for local lookups while preserving global search. No proprietary geocoding.
- Street infrastructure scoring via Overpass API (800m radius) We query within an 800m circle for:
highway=crossing nodes → crossing safety footway=sidewalk and highway=footway ways → sidewalk coverage highway=primary/secondary/tertiary/residential/living_street → network topology Way attributes: lanes, width, surface, maxspeed, lit, sidewalk, cycleway
Four sub-metrics from this graph:
Intersection density (nodes with degree >= 3 per km2) Average block length (total street length / intersection count) Network density (total street km per km2) Dead-end ratio (degree-1 nodes penalize walkability)
These combine into the Network Design component (35% of the total score). 3. 15-minute city scoring via Overpass API (1,200m radius) Service reachability on foot, scored by nearest distance (<=400m = 100pts, <=800m = 75pts, <=1,200m = 50pts):
Grocery: shop=supermarket/convenience/greengrocer Healthcare: amenity=pharmacy/clinic/hospital Education: amenity=school/kindergarten/library Recreation: leisure=park/playground/sports_centre Transit: public_transport=stop_position/platform, highway=bus_stop, railway=station/tram_stop/subway_entrance Dining: amenity=restaurant/cafe/fast_food
This feeds the Accessibility component (25% of total score) and a separate 15-Minute City Score.
- Map rendering via Leaflet + OSM tiles Scored infrastructure overlaid on OSM base tiles. What’s missing, and what would help We’re explicit in the UI about what we can and can’t measure:
✓ Crossings exist and where ✓ Lit / not lit (where tagged) ✓ Service accessibility via POIs ✗ Pavement condition ✗ Sidewalk obstructions (vendors, parked bikes) ✗ Crossing quality (marked, signalled, raised), sparse outside Europe/North America
The most useful contributions for Southeast Asian cities: sidewalk=, crossing=marked/uncontrolled/traffic_signals, and lit= on way segments. These tags directly change scores for real addresses. Nimman Road would improve immediately with accurate crossing nodes added.
The project
SafeStreets is live at safestreets.streetsandcommons.com. Built by Streets & Commons, a civic tech initiative based out SEA If you’re mapping in SE Asia and want to see a specific street analysed, or if you work on pedestrian tagging schema, I’d love to hear from you in the comments
Discussion
Comment from solenoid jam on 11 March 2026 at 01:12
Hi. I’m currently traveling in Southeast Asia. I’m reviewing some maps and am curious about traffic in this region. The first thing I see is that pedestrians have minimal status on the road. This is especially noticeable in Vietnam, where motorbikes don’t yield to pedestrians and instead honk to get them out of the way. Furthermore, there are often no sidewalks along the roads, and if they exist, they’re poorly constructed or cars are parked on them. Even if you’re walking on the sidewalk, you have to step onto the roadway to avoid obstacles. It’s dangerous. I’d also like to point out that developing pedestrian infrastructure in cities is impossible without developing public transportation, such as buses. If there are no bus routes, people will ride motorbikes, and motorbikes don’t need sidewalks or pedestrian crossings. Perhaps the tool you’ve proposed isn’t very suitable for Asian conditions, as it’s based on the Western model of transport infrastructure. I also want to point out that I often notice that OSM tagging schemes are good for Western countries, but they’re not adapted to Eastern culture. For example, marketplaces in Asia have much more internal differences than in Western countries, and more tagging schemes should be developed for them. I was also surprised by narrow roads that are suitable for pedestrians and bikes, but not cars. OSM doesn’t have precise designations for such roads, since they aren’t pedestrian paths or service roads. I think these issues need to be addressed more locally and comprehensively, to utilize OSM’s flexibility to adequately represent reality on maps.