StreetIQ: AI-Powered Road Intelligence for Safer Navigation

What I Built

StreetIQ is an AI-powered road intelligence and navigation platform designed to make roads safer by detecting and mapping hazards such as potholes, cracks, waterlogging, and debris.

Road hazards are often reported late, inconsistently, or not at all. StreetIQ aims to turn these scattered observations into actionable road intelligence.

The platform combines computer vision with real-time mapping and community-driven reporting. Its key features include:

• AI-powered hazard detection: A custom YOLOv8 model detects road hazards, with TensorFlow.js enabling browser-based inference.
• Real-time hazard mapping: Displays road hazards on an interactive map using OpenStreetMap-based mapping tools.
• Navigation and warnings: Integrates routing and location services to help users understand road conditions along their routes.
• Crowdsourced reporting: Allows users to contribute hazard reports, including voice-assisted reporting.
• Hazard verification and updates: Supports verification workflows and real-time updates to improve the usefulness of road intelligence.
• Road intelligence for authorities: The long-term goal is to help municipalities and road maintenance teams identify problem areas and prioritize repairs.

Who could pay for it?

Potential customers include municipal authorities, logistics and fleet operators, navigation providers, and road maintenance organizations. StreetIQ could evolve into a road-condition intelligence service that helps these organizations reduce vehicle damage, improve route planning, and prioritize infrastructure maintenance.

The project is still evolving, and these are potential commercial applications rather than claims of existing paying customers.

Demo

Watch the StreetIQ demo:

https://www.youtube.com/watch?v=lJVnyaPadkw

Code

GitHub repository: https://github.com/adityaverma9777/StreetIQ

The repository contains the implementation and development work behind StreetIQ. Contributions, feedback, and improvements to its detection pipeline, mapping, and reporting workflows are welcome.

How I Built It

I built StreetIQ using an open-source computer-vision approach combined with a web-based mapping and real-time backend.

Technology stack:

• AI and computer vision: Custom YOLOv8 model with TensorFlow.js for browser-based road-hazard detection.
• Frontend: React, Vite, and React-Leaflet.
• Backend: Node.js, Express, and TypeScript.
• Database and geospatial data: Supabase PostgreSQL with PostGIS.
• Real-time communication: WebSockets and Supabase Realtime.
• Maps and routing: OpenStreetMap, OSRM, Photon, and Komoot geocoding services.

The core detection workflow uses a YOLOv8 model that can run in the browser through TensorFlow.js. This makes it possible to perform visual inference without sending every image frame to a remote inference server.

Detected hazards can then be connected to the wider platform through location-aware reporting, geospatial storage, mapping, and real-time updates.

The engineering challenge was not just detecting a pothole in an image. It was connecting detection with geographic context, useful reporting, navigation, and a system that can eventually support large-scale road-condition monitoring.

Why Does Open Innovation Matter?

Road safety is a problem that affects everyone,