SUPPLY CHAIN Calculator have beendeveloped.the existing Lamb Value3D imaging “Those are valuable tools for – and more profitable –processors, helping themmake more sophisticatedput to the test use of the data from new measurement devices,” A3D imaging tool combiningweight, fat and muscle, so the plan now is Graham said. off‑the‑shelf cameras with for longer‑term deployment in a feedlot.” Collaboration artificial intelligence has Research will also continue to The collaborative nature successfully estimated P8 use computed tomography (CT) of ALMTech has been its fat and muscle score of live cattle measurements of carcases, so the tool greatest strength, with a at a NSW feedlot. can be used to estimate lean meat yield partnership of 19 processors, of live cattle at feedlot exit. technology providers, The feedlot testing marks the first universities, government real‑world application of the 3D imaging How it works departments and research tool, developed as a collaboration Off‑the‑shelf red–green–blue and development between the NSW Department of Primary (RGB)‑scanning cameras (similar to an companies, including Industries and University of Technology Xbox camera) take 3D images of an MLA, from across the Sydney (UTS), supported by MLA. animal as it passes through a chute. three industries pledging This ability to automate live animal Using P8 fat and muscle scores obtained $6.7 million. assessment will reduce subjectivity and by ultrasound from expert assessors, the This is in addition to the increase confidence in performance and team used machine learning – a branch $4.8 million in federal Rural carcase yield predictions. of artificial intelligence – to ‘teach’ their Research and Development A goal of Advanced Livestock software to interpret 3D shapes into for Profit program funding. Measurement Technologies (ALMTech) muscle and fat deposits, and produce These partnerships played is to develop lean meat yield objective scores related to these traits. a role in ensuring new measurement technologies for use on “If a shape has not been previously seen, technologies are practical. live animals and carcases. the system can request some information “Coming up with the fancy Researcher Alen Alempijevic, from the for it to expand what it has learnt so far – technology is really only Centre for Autonomous Systems at UTS, in the same way a person builds on their about 10% of the job,” said the results were “very promising”. experience,” Alen said. Graham said. “Our goal was to deploy the 3D imaging There is potential for the information “Most of the work in tool at a feedlot and have it automatically produced by this technology to be developing these report P8, hip height and muscle score,” fed back into the seedstock sector for technologies happens when he said. Estimated Breeding Values to predict they’re deployed in the lean meat yield at slaughter and real world. Cameras, attached to a custom‑designed pave the way for carcase sorting or portable chute at the feedlot, reported P8 value‑based marketing. “It’s not until you start values within 1.5mm of scanned values. integrating technology into “We didn’t push the envelope too far in Alen Alempijevic the processing environment terms of diversity of cattle, variation of live E: Alen.Alempijevic@uts.edu.au where you’ve got variations in light, chilling regimens, times after death, dehydration profiles and so on, that you can see where the problems are.” Even in live animals, the variable amount of dirt in the hide can affect the performance of devices, so the seamless deployment of these technologies into the real world has been crucial. ■ Graham Gardner E: g.gardner@murdoch. edu.au UTS researcher Alen Alempijevic trialling the 3D imaging tool on Angus cattle at a Northern Tablelands feedlot. 36