FloodDetect
See the water before it closes the road.
FloodDetect turns the public traffic and CCTV cameras already on your roads into flood sensors. It maps water on the roadway, estimates how deep it is, and alerts your city or county so you can close roads, send crews and fix the drains that keep failing.
Where the water is, not just that it’s there
Gauges and low-cost sensors measure a single point, usually in a drain or channel rather than on the road. 311 calls and crowdsourced reports arrive after the fact. Cameras are already installed and maintained on most corridors, and they show how far flooding has spread across the lanes.
The clip is a real night: a public OHGO traffic camera at Cleveland Ave and Innis Rd in Columbus, Ohio, on August 20, 2026. FloodDetect read one frame every 15 minutes, mapped the water in each, and charted it as it rose twice and drained by morning.
- Uses existing roadway cameras, so there are no new sensors to deploy
- Shows where flooding is, how large it is and, where possible, how deep
- Catches water early and pushes alerts before anyone calls it in
- Scales from one intersection to every flood-prone corridor
One night at Cleveland Ave and Innis Rd
- Rain forecast for the day at the camera location
- 4.2 in
- Of the camera view flooded at the peak, 3:46 AM
- 83%
- The second surge kept more than half the view flooded
- 3 hr+
- Frames read that day, one every 15 minutes, with no new hardware
- 95
Detect, then respond
From camera frame to road closure
An alert only matters if someone acts on it. When flooding crosses the thresholds you set, FloodDetect tells the people who can do something about it: traffic operations, public works, first responders and the engineers who plan drainage.
Each alert carries the camera, the location, a frame with the water outlined and how much of the road is covered, so a dispatcher can decide in seconds whether to close the road.
- Close and reroute: barricade the approaches, update message signs and traveler information before drivers reach the water
- Send resources where the water is: pumps, crews and water rescue go to the intersections that are actually flooded
- Fix the infrastructure: intersections that flood storm after storm point to clogged inlets, undersized culverts and drainage to repair
Who gets the alert
One detection, every team that responds
FloodDetect watches every camera you already have, so each department sees the same picture at the same time.
- 01
Traffic and emergency operations
Close lanes or roads, post detours and update 511 and message signs as soon as water covers the roadway. Stage water rescue near the worst flooding, route units around it, and reopen when it drains.
- 02
Public works
Dispatch crews and pumps to the flooded locations first, and clear the inlets the cameras show backing up.
- 03
Engineering and planning
Use the flood history at each camera, with time, extent and rainfall, to rank drainage projects and back up grant applications.
How FloodDetect works
- 01
Camera calibration
Each camera is calibrated and its road plane modeled, so pixels map to real-world ground positions.
- 02
Water segmentation
Deep learning models separate standing water from wet pavement and reflections, frame by frame.
- 03
Inundation mapping
The flooded area on the roadway or shoulder is measured and tracked as it grows or recedes.
- 04
Depth estimation
Curbs, lane markings and other known references give a relative water depth where the view allows.
- 05
Threshold alerts
You set the thresholds, such as 2 inches of water or a lane blocked. Crossing one alerts your traffic management center, first responders or message signs.
- 06
Dashboard and history
Every event is logged with time, location, footprint and depth, so you can find hotspots and plan drainage work.
Can’t wait? Contact us today.
The ClearObject team of AI experts is here to help you get started.
Talk to an expert