When a swimmer sinks towards the bottom of the pool, the new system sends an alarm signal to a poolside monitoring station and a lifeguards paper. In trials at a pool in Ancenis, near Nantes, it saved a life within just a few months, says Alistair MeQuade, a spokesman for its maker, Poseidon Technologies.
Poseidon keeps watch through a network of underwater and overheard video cameras. AI software analyses the images to work out swimmers trajectories. To do this reliably, it has to tell the difference between a swimmer and the shadow of someone being cast onto the bottom or side of the pool. The underwater environment is a very dynamic one, with many shadows and reflections dancing around. Says McQuade.
The software does this by projecting a shape in its field of view onto an image of the far wall of the pool. It does the same with an image from another camera viewing the shape from a different angle. If the two projections are in the same position, the shape is identified as a shadow and is ignored. But if they are different, the shape is a swimmer and so the system follows its trajectory.
To pick out potential drowning victims, anyone in the water who starts to descend slowly is added to the softwares pre-alert list, says McQuade. Swimmers who then stay immobile on the pool bottom for 5 seconds or more are considered in danger of drowning. Poseidon double-checks that the image really is of a swimmer, not a shadow, by seeing whether it obscures the pools floor texture when viewed from overhead. If so, it alerts the lifeguard, showing the swimmers location on a poolside screen.
【职称英语阅读理解练习题】相关文章:
最新
2016-03-01
2016-03-01
2016-03-01
2016-03-01
2016-03-01
2016-03-01