IT SOUNDS straightforward: give staff a second chance to examine images, and you could cut error rates in medical image analysis or improve airport baggage screening. But second viewings don鈥檛 always work, because people might focus on the same areas they looked at first time around, says of the FX Palo Alto Laboratory in California.
Now Qvarfordt and colleagues think they have found a way to uncover detail that may have been missed. They devised a system that uses eye-tracking technology to identify which areas within an image have been checked by the viewer. On a second viewing, those areas are blacked out.
The system was tested by asking volunteers to identify specific objects in images containing 300 confetti-like shapes. On a second viewing, those presented with the partly blacked-out images found significantly more shapes than people simply shown the original a second time.
Advertisement
鈥淚 think the technique has potential for augmenting inspection systems,鈥� says at Clemson University in South Carolina. Qvarfordt presented the work at the in Austin, Texas, last week.
