California's Camera Network Spots Wildfires Before Anyone Calls 911
Quick Summary
ALERTCalifornia, run by UC San Diego with CAL FIRE, watches more than 1,200 cameras around the clock with AI that looks for smoke. In a recent 12-month period it flagged about 3,600 wildfire incidents, and more than half of those detections came before any 911 call.
The Use Case
A fire in remote terrain can burn for a long time before anyone notices and calls it in, and those minutes decide whether it stays small. The camera network already existed for human monitoring, but nobody can watch 1,200 feeds at once. AI scans every feed continuously for smoke plumes and early fire signatures. On a likely detection it alerts fire agency staff, who confirm on the live feed and dispatch. The same cameras are used for floods, storms and debris flows, and the public can view them.
Real Example
- At 2:14 pm a camera on a ridge picks up a faint smoke column
- The model flags it and pings the CAL FIRE duty officer with the camera feed and location
- The officer confirms it on two cameras within three minutes
- An engine and air attack are dispatched before the first 911 call arrives at 2:31
- The fire is contained at under an acre
- The detection and response times are logged for the annual review
Why It Matters
- 1,200+ cameras monitored continuously by AI
- About 3,600 wildfire incidents flagged in 12 months
- More than 50% of detections came before any 911 call