CivicLens is a public infrastructure monitoring platform for Indore. A citizen uploads a photo of a pothole, road crack, garbage pile or broken streetlight and selects the location. A YOLO model we trained detects the problem, and a transparent scoring formula assigns a priority from 0 to 100, with a breakdown showing why.

Most existing tools stop at detection. CivicLens focuses on follow-through. Reports of the same problem within 20 meters merge into one ticket with a report count. Each ticket has a repair deadline based on its priority and is flagged when overdue. To close a ticket, staff upload an after photo and the model checks that the problem is gone. Tickets can be reopened, and a new issue near a spot repaired in the last 90 days is flagged as a repeat repair. The dashboard has a map, heatmap, issue queue, analytics, and a complaint letter addressed to the correct department.

The model was trained for 50 epochs on a Roboflow dataset. It is strongest on potholes, cracks and garbage, and weaker on drains.