And finally la piece de resistance, although any vegetarians out there may not wish to read further. The Danish Meat Research Institute has developed a system using neural networks to grade the meat percentages of pig carcasses to gain a more consistent result than is possible with traditional hand-held devices. Introduced in 1991 at its Classification Centre, the system grades the pigs at a rate of one every 10 seconds by penetrating each carcass in four different areas with a total of nine sensor probes. These determine whether the flesh is fat or lean meat and how thick it is by measuring the varying amount of light the two reflect onto the receiver. The neural networks are used to determine the meat content in place of the original rule-based software. The Institute is now working to perfect its latest project that began two years ago, this time for beef carcass grading from grey scale video images. Mmmm, time for lunch and a tasty ham sandwich… – Abigail Waraker
