Curb Cutagenda packets, zoning staff reports, public comment, 311 requests, survey free

Street Audits

Photographing Fifty Curb Ramps: A Volunteer Method for Street Audits and Vision-Tool Errors

By the Curb Cut editors / / 5 min read

Short answer. A useful curb ramp photo audit needs consistent viewpoints, location records, repeatable observations, and a human review process. Vision tools can help organize images and flag possible barriers, but volunteers must independently assess slopes, transitions, surface conditions, obstructions, and details that photographs cannot establish.

Key points

  • Use a consistent photographic protocol so observations from different volunteers can be compared.
  • Record location, direction, date, and visible conditions separately from AI-generated descriptions.
  • Vision tools can miss tactile surfaces, subtle level changes, temporary obstructions, and drainage problems.
  • A fifty-ramp audit is a manageable sample for organizing a neighborhood-scale inventory, not proof of conditions across an entire city.
  • Send documented findings to the public agency responsible for maintaining the location.

A Fifty-Ramp Audit Needs a Defined Route

A volunteer curb ramp audit begins with a route, not a camera. Choose a defined neighborhood area, such as the intersections along five consecutive blocks, and identify approximately fifty ramp locations before photography begins. A location may have two or more ramps, so distinguish an intersection from an individual ramp in the inventory. This prevents a common counting problem: recording one corner as a complete intersection and overlooking the crossing on the opposite side.

Create a simple inventory with a unique identifier for each ramp, its street intersection, the side of the street, and its relationship to the crossing. For example, an identifier such as R-024 can refer to the northeast corner of a named intersection. Keep the location record separate from the photograph filename. That separation makes it possible to replace an image, correct a description, or revisit a location without losing the history of the observation.

  • Define the geographic boundary and intended sample.
  • Assign one unique identifier to each ramp.
  • Record the intersection and corner orientation.
  • Mark locations that cannot be photographed safely.
  • Keep a route map alongside the inventory.

Use a Repeatable Photographing Method

Photographs are useful only when their viewpoints provide enough context. For each ramp, take an approach view showing the sidewalk and surrounding crossing, a closer view of the ramp surface, and a view that captures the transition toward the street. These three images answer different questions. The first shows the relationship to the pedestrian route, the second records visible surface details, and the third helps reviewers understand how the ramp meets the roadway.

Keep the camera orientation and distance reasonably consistent. Include the ramp identifier in the photo log, not necessarily in the image itself. Record whether parked vehicles, construction equipment, vegetation, shadows, or other conditions obstructed the view. A photograph taken from a safe standing position is preferable to one that requires a volunteer to enter traffic or stand in an unsafe location.

A photo audit is not a substitute for specialist measurements. A photograph may suggest a steep transition, uneven surface, or missing tactile warning, but perspective and camera angle can distort apparent dimensions. Record visible observations as observations. If a project requires measurements of slope, width, or other technical characteristics, identify the appropriate measuring method and the responsible accessibility professional or public agency.

Separate What Volunteers See From What Vision Tools Describe

Computer vision can assist with image organization by identifying likely curb ramps, grouping photographs by visible features, or producing preliminary descriptions. A language model connected to an image-analysis service might describe a photograph as showing a concrete ramp with a detectable warning surface. That description is a starting point for a human record, not a finding about accessibility.

Vision systems can misclassify tactile warning surfaces when their color resembles surrounding pavement. Shadows can look like level changes, while worn or dirty surfaces may disappear into the background. A model may also describe a ramp as unobstructed because it cannot recognize that a vehicle blocks the actual pedestrian approach outside the camera frame. Wet pavement, snow, temporary construction, and poor lighting introduce additional ambiguity.

Keep the original image, the tool's description, and the volunteer's observation in separate fields. A useful record might say that the model identified a tactile surface, while the volunteer recorded that the surface was partly obscured in the photograph. This difference is important. Combining the two descriptions into one apparently authoritative statement hides the uncertainty that should guide any later inspection.

  • Treat image descriptions as preliminary labels.
  • Flag low-light and obstructed photographs.
  • Preserve the original image alongside any generated description.
  • Use a human review field for disputed classifications.
  • Escalate measurement questions to qualified personnel.

Build an Inventory That Can Be Mapped and Corrected

A spreadsheet is usually sufficient for a small volunteer audit. Useful columns include ramp ID, intersection, corner, latitude, longitude, photograph references, visible surface condition, apparent obstruction, tactile surface visibility, and review status. Use controlled categories where possible. For example, an obstruction field might contain clear, partly obstructed, blocked, or unable to determine. Keep free-text notes for details that do not fit those categories.

A map can show where observations cluster and where information is missing. Display individual ramp locations rather than assigning one accessibility status to an entire intersection. A map marker should communicate that an observation exists, not that a location has received an official accessibility certification. If coordinates come from a phone, retain enough location information to recognize possible errors around closely spaced corners.

Before sharing the inventory, inspect duplicate photographs, misplaced coordinates, inconsistent identifiers, and blank fields. AI-assisted data cleaning can flag records with missing values or conflicting descriptions, but a volunteer should decide whether two entries refer to the same ramp. A fifty-location inventory can also be divided into review batches, making it easier to assign follow-up work without giving every volunteer access to the entire dataset.

Turn Photo Findings Into a Public Works Request

A useful audit has a clear destination. The responsible public works, transportation, or accessibility office can explain which information belongs in a maintenance request and which concerns require a formal assessment. Reporting a damaged surface, an obstruction, or a missing crossing connection is more actionable when the submission includes an intersection, ramp identifier, photograph, observation date, and a concise description of the visible problem.

Do not label an entire location compliant or noncompliant based on a volunteer photograph. A city may have separate processes for routine maintenance, capital improvements, and accessibility planning. Ask which process applies and whether the office maintains an existing inventory. If an official record differs from the volunteer map, preserve both records and identify the difference for follow-up.

The next step is to select one documented location with a clear visible concern, attach its identifying photograph and intersection details, and submit it through the appropriate municipal reporting channel. Ask the receiving office which department owns the issue, when it will be reviewed, and how residents can follow its progress.

Questions readers ask

How many photographs should volunteers take at each curb ramp?

Three consistent views provide a practical starting point: an approach view, a surface close-up, and a view of the street transition. Additional photographs are useful when an obstruction or unusual feature needs context.

Can a phone camera measure curb ramp slope?

A photograph alone cannot establish a reliable slope measurement. Use an appropriate measuring method and consult the relevant agency or accessibility professional for technical assessment.

What should happen when two volunteers describe the same ramp differently?

Preserve both observations and the associated photographs. Assign the record for a second human review instead of automatically selecting the description produced by an AI tool.

How can a neighborhood group keep its audit useful over time?

Retain the observation date and add a new record when conditions change. Separate historical photographs from current conditions so that an old obstruction is not presented as a current problem.

Related reading