Grant Award View - GA300639
Using machine learning with DXA as a technique to detect bones and...
Dual Energy X-ray Absorptiometry (DXA) is an imaging modality that can predict lamb carcass composition with incredible precision and accuracy at chain speed in multiple Australian abattoirs. In addition to composition information for individual lambs, an algorithm has been developed to predict the weight of almost all possible commercial cuts by using the DXA results with hot carcass weight. This cut weight prediction can be improved by virtually ‘sectioning’ the carcass DXA images into fore, saddle and hind sections, but this method requires the installation of an additional x-ray tube for stereoscopic image analysis. Through machine learning and deep neural network analysis, DXA images from a single x-ray source will be virtually sectioned by using the cutting lines informed by the second x-ray source as the training data.