Electronic lifeguard watches pools

Judgment: A system of cameras' success depends on its ability to distinguish between a potential drowning and regular swimming.

October 08, 2001|By Kevin Washington | Kevin Washington,SUN STAFF

The video is subtle, but chilling. There, on the floor of a public pool in France, a swimmer has become almost motionless - his lungs filling with water. The view shifts from one camera to the next.

What doesn't appear on the video, but remains central to the rescue of the 18-year-old man on the pool bottom is a network of video cameras, a computer running analyses of what the cameras are relaying and an alarm system with pagers that acted in concert one November day last year to alert lifeguards that Jean-Francois LeRoy was drowning.

Today, the young swimmer, who was seeing just how far he could swim underwater, owes his life to the Poseidon anti-drowning detection system that made its American debut late last month at public school pools in St. Cloud, Minn.

Created by Vision IQ, a Paris-based company, the drowning detection system centers on a sophisticated computer program that interprets what it sees through a network of cameras above and below a swimming pool.

Vision IQ President and General Manager Steve Bagby calls this all "smart vision technology" in which you're "equipping the computer to look at a scene and react to what it sees."

The sad statistics of drowning provide an ample opportunity for technology to assist humans. According to the Centers for Disease Control, on average six people drown in U.S. pools every day. And about four or five suffer injuries from nearly drowning for every person who dies.

At heart, the system acts as an aid to lifeguards who may not be able to see everything going on in a pool or who have lost vigilance, simply because they're human, says Bagby, who is based at the company's U.S. headquarters in Boston.

Detection begins with a network of several small globe style video cameras installed in water- and pool-chemical-proof housings placed in the walls of the pool. They cover the deep end of the pool while other video cameras are mounted overhead on the ceiling or poles to watch the shallow areas where people playing can obscure an underwater view of the bottom.

The cameras are patched into a computer - a Pentium III-class machine for every 10 cameras - that sees the video in real time and analyzes it for a swimmer's behavior in the water. When the program detects a motionless swimmer for more than 10 to 15 seconds - possibly drowning - an alarm will sound and a light goes off on the monitoring station. A pager carried by the lifeguard will beep and vibrate while showing coordinates of the potential victim in the pool on an LCD screen.

The system's secret is that the computer's vigilance never flags no matter how long it's on. Bagby notes that studies of human vigilance show that having someone watch constantly for an event that rarely occurs is a recipe for disaster. "At the beginning of the shift, vigilance is pretty good, but after the first half hour, you're in pretty bad territory," he says.

Take into account that many lifeguards are young, temporary staffers, paid minimum wage - making the situation less than ideal.

Michael Olsen of Jeff Ellis and Associates, an aquatic-safety consulting firm working with Vision IQ, says lifeguards have all kinds of distractions from dehydration to ogling people in swim suits. "Being a teen-ager, you get distracted."

The drowning-prevention technology has been in development since Vision IQ was founded in 1995, but it wasn't until 1998 that the speed of computers allowed the software to develop for completing the complex tasks of identifying a drowning swimmer.

Computer interpretation of data isn't based on experience as with human beings, but on mathematical equations. As human beings, we can recognize all manner of chairs, Bagby says. A computer, provided a specific description of a chair, will only recognize a chair that fits the specific description.

"In simplest terms, understanding the difference between a shadow or patch of light and a swimmer is a pretty complex task," Bagby says. To add to the complexity, the computer tries to discern the trajectory of a swimmer - for example, is he sinking slowly - to determine if something suspicious is occuring.

You can see a brief sequence of LeRoy's near-death experience video in the pool in Ancenis Loire-Atlantique at the Poseidon's Web site, www.poseidon-tech.com.

Bagby claims his system has a 90 percent detection rate - nearly twice that of lifeguards who have been studied. And he says a public pool might have a false alarm once every couple of weeks.

The system isn't designed for home pools in part because "most people who drown in home pools never intended to go into the water," Bagby says. He said fences and detection systems for people entering the water are more appropriate for such pools.

Poseidon also can warn of after-hours swimmers and it can alert a pool manager to unsafe conditions before a pool opens - most notably when turbidity is so high that a lifeguard might not be able to see to the bottom of the pool.

Baltimore Sun Articles
|
|
|
Please note the green-lined linked article text has been applied commercially without any involvement from our newsroom editors, reporters or any other editorial staff.