Methodology
How route checks work
Detectable Design reads a planned GPX route in your browser, matches it to available
road data, highlights intersections with crash history or other risk factors, and adds clothing color
guidance where local street-scene profiles overlap the route.
Route guidance is directional. It is not a crash prediction or safety guarantee.
Route first The page starts from the route you plan to run or ride, not from a city selector.
Attention index Crossings are scored from road context, scene complexity, and nearby crash history.
Color guidance The route card returns two high-vis colors and two regular clothing colors to start with.
What happens to a GPX file
- The GPX file is parsed locally in the browser.
- The route is sampled into points at regular distance intervals.
- Those points are matched to nearby road-network tiles.
- Matched road segments provide road type and available safety features.
- Nearby intersections are ranked and the top five are shown on the route card.
Signals used by the route check
- Road class: local streets, arterials, highways, and active paths.
- Intersection context: crossings, road complexity, and nearby route geometry.
- Crash history: federal or local crash records where available.
- Street-scene profiles: local backdrop and color guidance where imagery exists.
- Season and light condition: season-specific profiles are used where available; otherwise the route uses all-year guidance.
How the attention index is built
The attention index is designed to make the route card easy to share: one route-level
label plus a short list of places to notice. It intentionally avoids pretending to know
the exact risk of a future trip.
Intersection score
Local comparison
Each matched crossing is compared with other crossings in the same local context.
More complex crossings, or crossings near crash records, receive higher scores.
Route score
Average attention
The route score summarizes the crossings along the uploaded GPX file rather than
ranking the whole city.
Route labels
Readable output
High, moderate, light, and easy attention labels are based on how the route compares
with nearby intersections in the available dataset.
Top five points
Actionable scan
The route card limits the watch list to five intersections so the output stays
usable before a run or ride.
How color guidance fits in
Color guidance is now a route output, not the main navigation model. When the route
overlaps local street-scene profiles, the page weights those profiles by the matched
route segments and recommends colors that separate from the local backdrop.
High-vis colors
The route card shows two high-vis colors to start with. These are still usually the
strongest visibility choices when someone owns high-vis gear.
Regular clothing colors
The route card also shows two regular clothing colors for cases where someone is
choosing from clothing they already own.
Colors to avoid
Avoid colors are the colors most likely to blend into the matched local backdrop
profiles.
Fallback guidance
If the route has road coverage but not local street-scene profiles, the page uses
general visibility guidance instead of pretending the local color model is complete.
Current limitations
Coverage varies by data source
Road coverage and local street-scene coverage are separate. A route can have road
matching without local color profiles for every segment.
Crash data is historical
Crash records describe what has been recorded nearby. They do not predict what will
happen on a specific trip.
Lighting is incomplete
Low-light guidance uses available route and scene data, but comprehensive streetlight
coverage is not available everywhere yet.
Season is available where data supports it
Seasonal profiles are used where there is enough image metadata. Otherwise the route
card uses all-year profiles.
Visibility research behind the color guidance
The route tool uses these studies to shape the color component of the recommendation.
They support the general structure of the visibility model, not every constant in the
implementation.
-
Targoutzidis, A., et al. Selecting effective colors for high-visibility safety apparel.
Safety Science, 2021.
doi.org/10.1016/j.ssci.2020.104978
-
Sayer, J., & Mefford, M. The effect of color contrast on daytime and nighttime conspicuity of roadworker vests.
UMTRI, 2000.
hdl.handle.net/2027.42/49418
-
Sayer, J., & Buonarosa, M. The roles of garment design and scene complexity in the daytime conspicuity of high-visibility safety apparel.
Journal of Safety Research, 2008.
doi.org/10.1016/j.jsr.2007.12.004
-
Wood, J., Tyrrell, R., et al. Limitations in Drivers' Ability to Recognize Pedestrians at Night.
Human Factors, 2005.
doi.org/10.1518/001872005774859980
-
Babić, D., et al. Factors affecting pedestrian conspicuity at night: Analysis based on driver eye tracking.
Safety Science, 2021.
doi.org/10.1016/j.ssci.2021.105257
Bottom line
Detectable Design is an information tool for planning routes and choosing more visible
clothing. It should be used alongside normal judgment, local conditions, lights,
reflective details, and route planning.
Check a route