Real-Time Pothole Detection Using Deep Learning

Detect potholes. Improve roads. Make travel safer.

Upload a photo of a damaged road and DriveSafe Vision generates a pothole report complete with severity classification, GPS location, estimated dimensions and annotated image evidence — then tracks it until the road is repaired.

Severity is estimated by AI from the photo — an inspector confirms it on site.

City road with a pothole outlined by a detection bounding box labelled 98% confidence
High severity1 pothole · 98% confidence

How it works

Four steps from spotting a hazard to seeing it fixed.

  1. 01

    Upload an image

    Add a photo of the damaged road surface.

  2. 02

    Analysis runs

    The detector locates potholes and estimates their size.

  3. 03

    Submit the report

    Attach your location and send it to road maintenance.

  4. 04

    Track resolution

    Follow the status from review through to repair.

Key features

Built for citizens reporting hazards and the teams that fix them.

AI-assisted detection

Potholes are located and outlined on the image.

Severity classification

Low, Medium, High or Critical, based on size and count.

Automatic location capture

Browser geolocation, with manual address entry as a fallback.

Annotated evidence

Original and annotated images are stored with every report.

Status tracking

A clear timeline from submission to resolution.

Admin review workflow

Administrators triage, note and resolve every report.

What coordinated reporting achieves

Illustrative programme figures shown as sample data.

Reports submitted
12,480
Potholes detected
31,207
Reports resolved
9,860
Average response time
4.2 days

Help make your roads safer.

Create a free account and submit your first report in minutes.

Create Your Account