Particle.news

Bengaluru Engineer Builds AI That Maps Potholes to Contracts

A prototype uses dashcam, GPS, an accelerometer plus vision models to auto-generate location-tagged complaints listing tender numbers, responsible officers, photos, geocoordinates.

Overview

  • A Bengaluru software engineer, Gaurav Sen, demonstrated a proof‑of‑concept in mid‑August that detects potholes from dashcam, GPS and accelerometer data and classifies them with vision models.
  • Sen’s demo showed the system searching roughly 2,900 government contracts to match a detected defect to a tender, identify the contractor and surface the responsible officer for a ready‑to‑file complaint.
  • The developer says the app can produce a structured output — photo, geolocation, tender number and officer — in about four seconds and turned one drive that found 12 potholes into 12 complaint records.
  • The demonstration went viral after being shared on X and drew widespread public interest, with reports noting Bengaluru already operates official pothole‑reporting portals that accept geo‑tagged evidence.
  • Journalists and experts stress this is a prototype with no confirmed municipal adoption and cite open questions about detection accuracy, contract-to-road mapping, data access, privacy and how authorities would validate or act on mass automated complaints.