No account
Drivers can interpret a sign without creating a profile.
Applied AI / Mobility
Parky.AI
AI Parking Rule Interpreter
A photo becomes a practical answer to the question drivers actually care about: “Can I park here right now?”
Overview
Parking signs are simple until several rules apply at once.
A single sign stack can combine restrictions based on time, day of week, season, direction, permit status, vehicle type, payment, and temporary conditions.
Parky.AI transforms a captured parking sign into a clear conclusion for the current moment, then explains how the visible rules led to that answer in plain language. No account or device location is required.
The challenge
Recognition was only the first step.
Most parking tools display rules. The harder problem is deciding which rules apply to one space at one point in time.
Drivers can spend minutes reading layered regulations and still remain uncertain. Different municipalities use different layouts, symbols, arrows, languages, and combinations of conditions, making a universal answer significantly more difficult than extracting text from an image.
Deliver a useful answer faster than a person can read and mentally evaluate the complete sign stack.
The reasoning engine
From visual evidence to an explainable decision
Underlabs evaluated multiple recognition and reasoning approaches to balance accuracy, response speed, battery efficiency, and privacy.
- 01Capture
The driver photographs the complete set of relevant parking signs.
- 02Structure
The system identifies individual rules, symbols, times, dates, permits, and directional arrows.
- 03Evaluate
Every visible condition is checked against the current day and time before a conclusion is produced.
- 04Explain
The result includes natural-language reasoning so the driver can review which restrictions apply.
The solution
A conclusion the user can inspect, not a black-box answer.
Parky.AI was developed as an AI-powered parking interpretation platform for iOS and Android. It analyzes the sign stack, directional arrows, each individual rule, the current date, and the current time.
Rather than returning only a binary result, the application explains why parking is or is not permitted based on the visible information. When the image does not support a reliable conclusion, Parky can return “Not sure” instead of presenting uncertainty as permission.
- Combines overlapping permissions and restrictions
- Evaluates arrows, days, dates, times, and permit conditions
- Returns Yes, No, or Not sure
- Explains the conclusion in the user’s language
Product in use
One photo, every visible rule, one practical conclusion.
The mobile workflow keeps the camera, decision, and explanation together so the user can act without navigating a rule database.
Visit Parky.aiPrivacy-first design
Useful context without tracking the driver.
No GPS required
The conclusion uses the captured sign together with the current date and time.
Images not retained
Uploaded parking-sign images are not kept after processing.
Worldwide interpretation
Built around Montréal complexity, extended to parking signs worldwide.
Parky evolved from Montréal’s layered parking system into a platform that works across municipalities and supports parking signs in 85 languages. The explanation is returned in the user’s language, helping residents and travelers understand the visible rules.
News and media
A practical AI product that earned broad attention.
Parky has been featured across television, radio, and digital publications in Montréal and Canada.
Result
AI that simplifies real-world decision making.
Parky.AI moves beyond image recognition by evaluating structured regulations and showing its reasoning. The result is a faster, clearer parking experience for everyday drivers.
- ImmediateA practical parking conclusion for the current moment
- ExplainableRule-by-rule reasoning instead of an unsupported answer
- PrivateNo account or device location required
- WorldwideParking-sign support across languages and municipalities
Project facts
ProductParky.AI
PlatformsiOS and Android
CapabilitiesComputer vision, rule interpretation, reasoning engine, multilingual explanations
PrinciplesExplainable AI, privacy by design, practical decision support
Parky interprets the visible information in the submitted image. Temporary notices, nearby signs, curb markings, or other conditions outside the frame may still apply.
Build useful applied AI
Turn a complex real-world decision into a clear product experience.
Share the workflow, evidence, rules, and uncertainty your users need help navigating.
Discuss your AI product