In Norway it's compulsory to classify all types of carcasses. The classification is a quality assessment of the carcass within a standardized framework. Norway uses the same classification system as in the EU; the EUROP system. Classification of sheep is a subjective assessment done by trained and calibrated classifiers. The assessment of the classifiers is inspected systematically throughout the year.
The classification system for sheep consists of three factors: category; selection in groups after age and sex of the animals; conformation- and fat groups. The classification results and carcass weight are the basis for payment to farmers and prices to the market. Information from the classification is also used to optimise cutting and production processes in the cutting departments.
The MeatCrafter project lasted from 2017 to 2021. The aim was to develop, test and verifying an automatic, objective, fast working, cheap, non-contact classification of lamb and sheep. An additional goal was to avoid potential classification differences between classifiers, which may occur with manual classification. We also had an expectation that it would be possible with MeatCrafter to predict meat-, fat -, and bone percentages with higher precision.
MeatCrafter is based on 3D imaging and reflections of near infrared (NIR) spectroscopy. In 2017, SINTEF carried out initial experiments with NIR on a few cuts of meat. The conclusion was that NIR can be used to predict fat content with high precision.
In January 2018, the prototypes were tested on a selection of carcasses in Animalia's pilot plant. These carcasses were also CT-scanned and cut and deboned so that the body composition, i.e. the amount of meat, fat and bone, was determined in detail. Based on these results, it was decided to proceed with the trial on a larger scale. Furthermore, it was concluded that the resolution of the 3D camera was not crucial, therefore, cheap regular cameras were introduced.
This prototype was tested on approximately 5000 individuals in the abattoir Fatland Oslo during the fall of 2018 and spring of 2019. 52 of the individuals were CT-scanned and dissected at Animalias pilot plant. These were the first trials where MeatCrafter was deployed along a commercial slaughter line. Significant challenges regarding data collection were identified. In addition, the MeatCrafter equipment had to be removed when pigs were slaughtered on the same line. Data from individuals where the MeatCrafter instrument and data collection worked as intended, including individuals that were CT-scanned and slaughtered at the pilot plant, confirmed that
MeatCrafter has the potential to satisfactory predict EUROP class and fat group, as well as provide a good prediction of the amount of meat, fat, and bone.
In august 2019 we moved the project from Fatland Oslo to the abattoirs Fatland Jæren and Nortura Forus. This was the first time the MeatCrafter was permanently fixed to the slaughterline. We had to establish an external data warehouse for all the recorded data, as these two abattoirs are among the largest in Norway. The MeatCrafter generates 10 Mb of data per carcass.
Since 2019 there have been technical challenges with which we have had to work on up until the project ended in 2021. The MeatCrafter equipment requires that each carcass passes the instruments at a constant speed and the back facing the instrument. To make this possible, we had to substitute part of the line. A temporary solution to ensure correct positioning of the carcasses was developed and tested in august 2021. The NIR measurements were influenced by background and reflected light; however, a solution to avoid this issue has been implemented at Fatland Jæren.
Before MeatCrafter can efficiently be used for automatic classification of lamb, some work remains. An automatic system demands traceability at carcass level along the entire slaughterline. Furthermore, a correct link between data and carcass must be ensured. Also, appropriate solutions for correct carcass positioning and control of light reflections, must be improved. The abovementioned is crucial for NIR light measurements to provide standardized and consistent results. The consortium plan to address these remaining challenges in a new project application.
Prosjektet mål var å automatisere/ objektivisere klassifiseringsprosessen av sau og lam. Vi har i prosjektperioden ikke nådd målet. Vi har utviklet registreringsinstrumenter som er relevante for formålet. For å nå det endelige målet må underliggende utfordringer løses. Sjøl om prosjektet avsluttes, ønsker prosjektdeltagerne å løse disse utfordringene for å kunne ta i bruk Meatcrafter. Dette må gjøres gjennom et nytt prosjekt.
Klassifisering av slakt bør være en objektiv vurdering av slaktet som er basert på sammensetningen av kjøtt, fett, og bein. Per i dag er klassifiseringen det nest viktigste grunnlaget for pris til bonde og industri, etter slaktevekt. I Norge er EUROP-skalaen brukt for å klassifisere slakt - det er en manuell og subjektiv metode.
MeatCrafter: Automatisk, rask, rimelig og ikke-kontakt klassifisering av sau/lam - en norsk innovasjon med et internasjonalt marked.
Prosjektet skal utvikle et automatisk, objektivt og ikke-kontakt klassifiseringsinstrument for lamme- og saueslakt basert på optisk 3D-avbildning, spektroskopi og multi-variat dataanalyse. Skrottene vil klassifiseres på grunnlag av biometri, tredimensjonal form, farge, mengde overflatefett og fettlagets tykkelse. Prosjektet skal utvikle instrumentet og algoritmer for beregning av EUROP-klassifisering, inkludert videre registrering i slakteterminaler og databaser, og det skal oppfylle dagens regelverk. For å øke robustheten og objektiviteten av målingene skal instrumentet ikke kalibreres mot manuell klassifisering; det skal kalibreres mot utbytte bestemt både fra nedskjæring og CT-skanning av lammeskrotter, jfr. disseksjon av gris.
Objektive målinger med god presisjon og repeterbarhet vil gi en ny forbedret «standard» og en presis og robust verdivurdering av slaktene både til bonde og til industri. Prosjektet er godt forankret i kjøttbransjens styringsorganer og tematiske FoU-prioriteringer for mer effektiv produksjon.