Pilot program · Escazú, Costa Rica
How good are our streets?
Cleaner, more livable cities for everyone start with streets you can measure.
Want this for your city? Request it
Drag to pan. Hold ⌘ or Ctrl to zoom. Tap a segment to open it in the platform.
- 1 street camera-observed (2 sessions reviewed) · <0.1% of network · pending field verification
- 1 community contribution · unverified
Lowest camera scores
Cleaner, more livable cities for all.
We want cities that are cleaner, calmer, and more livable for everyone who moves through them. That does not begin with a bigger plan. It begins with an honest look at the street.
Street quality is not just pavement. It is where infrastructure, daily life, and community intersect, and its condition quietly decides who a city serves and who it leaves behind.
StreetLens is the instrument for that. The method is open and the same rubric applies in any city, so anyone can read a street the same way and check the result. Today the pilot runs on demo data over real geometry, and fieldwork in Escazú begins in August 2026.
- Measure what matters
- Accessibility, drainage, shade, and bike infrastructure, scored segment by segment. A city that goes unmeasured cannot be made fair.
- Implement, don't just propose
- We would rather measure one real street than publish another plan. Every reading is built to be concrete and actionable, and the data will grow into a public record other cities can learn from.
- Grassroots and systemic
- Residents can score the streets they walk every day, while open data gives policymakers a shared, checkable picture. That is how better, cleaner, more livable cities get built for everyone.
Four lenses on every street.
Walkability breaks down into parts you can measure. Every segment is scored on four independent rubrics, the same way in any city. Two of them are climate resilience: shade tracks heat exposure, and drainage tracks flood risk. Each ramp is colorblind-safe and paired with a line-width channel, and every legend shows its bins.
Accessibility
Curb ramps, tactile paving, and safe crossings. Whether everyone can actually use the sidewalk.
Drainage
Grates, cross slope, and ponding history. From flood-prone to well-drained.
Shade
Tree canopy and cover, and the heat they hold off. From exposed pavement to full canopy.
Bike
Bike lanes and separation from traffic. From none to protected.
Accountability begins with a number.
You cannot budget, fix, or answer for what you never measured. A complete street starts as a measured one, and most cities cannot say which sidewalks fail, or where.
Accessibility, drainage, and shade are almost never measured segment by segment, so they never enter a budget or a work order.1Ley 7600, Costa Rica's equal-opportunity law, has required accessible sidewalks since 1996. No municipality yet publishes a compliance map. The cost of not knowing is not abstract. It is counted in people.2COSEVI, the national road-safety council, publishes Costa Rica's annual road-safety counts.
- 1,601
- pedestrians run over in Costa Rica in 2024
- 30
- years of Ley 7600, still with no municipal compliance map
COSEVI road-safety data.
Ley 7600 took effect in 1996.
Escazú first. Then every corridor that will have us.
Every instrument needs a first street. Ours is San Antonio de Escazú, mapped segment by segment over real geometry. Field collection begins in August 2026, corridor by corridor.
The pilot corridor, mapped in full.
San Antonio de Escazú over real OpenStreetMap geometry. The rest of the canton follows, on the same rubric.
Source: StreetLens over OpenStreetMap geometry. Demo scores until fieldwork begins August 2026.
No field audits published yet. The canton network is live on the map; camera walks and community reports are the signals that move today.
- 1 street camera-observed (2 sessions reviewed) · <0.1% of network · pending field verification
- 1 community contribution · unverified
This is a pilot, and a beginning.
We walk the street, then we score it.
Each segment is audited against an established rubric. Nothing is inferred from a desk.
An auditor walks the segment, records each item on site, and the score is computed from those observations.3MAPS-Mini is a validated microscale audit of pedestrian streetscapes: sidewalks, crossings, and buffers. The same instrument runs on any street network, and every number traces back to something a person saw and wrote down.4LANAMME-UCR is Costa Rica's national road-materials laboratory at the University of Costa Rica, the source for pavement and condition standards.
MAPS-Mini
A validated microscale audit of pedestrian streetscapes: sidewalks, crossings, and buffers.
LANAMME-UCR
Costa Rica's national road-materials laboratory at the University of Costa Rica, for pavement and condition standards.
Ley 7600
Costa Rica's equal-opportunity law, which sets the accessibility minimums a sidewalk must meet.
Field audit
Every score traces to an on-site observation and, in time, a photo.
What the four lenses actually read.
A street's right-of-way in section, with the sidewalk, buffer, and carriageway each measured against the rubric above.
Source: StreetLens drawing over Reglamento a la Ley 7600 (Decreto 26831-MP) sidewalk minimums.
Calle Central, San Antonio
73
Overall
- Accessibility
- 41
- Drainage
- 68
- Shade
- 55
- Bike
- 30
How one segment becomes four scores.
The same segment in plan, with the audit rubric, score bins, width channel, and community casing that resolve to the readout above.
Source: StreetLens scoring schematic. Width bins per Reglamento a la Ley 7600 (Decreto 26831-MP).
How the data is collected
Two ways in, one reviewed dataset.
Field audits
A trained auditor walks a corridor and records the rubric on site. This is the source of truth.
Community reports
Anyone can add a street we missed or flag a correction from the map. Every submission is reviewed before it changes anything public.
The engine ahead
Every audit trains the next one.
Field audits remain the source of truth, and they stay that way. Camera walks already propose scores on the map: each session is reviewed before anything goes public, and a person still verifies what it reports. Those walks also become labeled training data so future field audits score faster. A camera proposal is not a field audit until one is published.
Field audits today. A verified model tomorrow.
Every walked corridor is photographed, and those images become training data for a computer-vision pipeline that a person always checks.
Live today: camera proposals on the map, admin-reviewed. Field audits remain the published source of truth.
Open data, or it didn't happen.
StreetLens is not a black box. It leans on established rubrics and open datasets, every score ships with its formula, and the code is public.
- MAPS-Mini
- Microscale Audit of Pedestrian Streetscapes.
- LANAMME-UCR
- National road-materials laboratory, University of Costa Rica.
- Ley 7600
- Equal Opportunities for Persons with Disabilities Act.
- OpenStreetMap
- Street geometry, under ODbL 1.0.
- COSEVI
- National Road Safety Council road-safety data.
Open source
The whole platform is public on GitHub: the code, the rubric, and the scoring method, open for anyone to read, run, or fork.
Open data
Every segment exports with its geometry and its score formula.
From one corridor to an open standard.
The pilot is step one. The instrument is built to scale, from a single corridor to a whole canton, across cantons, and out to anyone who needs the data.
- 01Now
Pilot corridor
San Antonio de Escazú, mapped over real geometry. Field collection begins August 2026.
- 02Planned
Full canton
Every corridor in Escazú, audited on the same rubric.
- 03Planned
Canton comparison
Neighboring cantons measured the same way, so results compare directly.
- 04Planned
Open API and data
Governments and researchers query the scored network directly.
Honest answers.
Is this real data?
The street network is real OpenStreetMap geometry. Demo audit scores are off by default; what you see colored today is camera-observed or community-reported work, each reviewed before it appears. Field audits will replace demo scores corridor by corridor once published.
How are scores computed?
Each layer is scored 0 to 100 from a rubric of observed items, where higher is better. The map pairs a colorblind-safe color ramp with a line-width channel, and the legend always shows the value bins. Every score ships with its formula.
Does StreetLens use machine learning or computer vision?
Yes, for proposals — not for published field audits. Camera walks already produce proposed scores on the map; a reviewer approves each session before it appears, and those readings stay separate from field audits until one is published. Field auditors remain the source of truth, and a person always verifies what the camera reports.
Can my city use this?
That is the goal. The rubric, the app, and the data format are open and portable, built to run on any street network. Today only the Escazú pilot is live. If you want to bring StreetLens to your city, reach out through the project on GitHub.
When does fieldwork start?
Field collection in San Antonio de Escazú is planned for August 2026. Until then the pilot runs on demo data over the real corridor.
Can I contribute?
Yes. You can add a street we missed or flag a correction from the map. A reviewer checks every submission before anything on the public map changes.
Who is behind this?
StreetLens is a project of CUSP, the Collective for Urbanism and Sustainable Progress, led by Gianluca Fonseca. It is open source and built in the open.
The call
Measure what the city forgot to.
Open the live map, or add a street you know. Every score ships with its formula.