Microwave array system for measuring fat and muscle depth of beef striploin and cube roll (Stage 1 – Pre-alpha prototype)

This project investigated whether microwave scanning technology could provide a more accurate, non-invasive method for measuring subcutaneous fat depth and help guide trimming to meet market specifications.

To test this, 100 beef striploins were scanned using a custom microwave system with four antennas operating over a wide frequency range. Each loin was sliced into steaks, and fat depth was manually measured at standard points using imaging software. These measurements were then compared against the microwave signal data using advanced machine learning models designed to optimise prediction accuracy. 

The results showed that the microwave system, combined with ensemble machine learning, achieved strong predictive performance. Accuracy improved when fat measurements were averaged across multiple points, with overall prediction error reduced to within approximately ±3 mm of actual fat depth. Antennas two and three consistently performed best, and the system produced results comparable to or better than traditional manual scoring methods. 

This project confirmed the technical feasibility of microwave scanning for objective fat depth measurement.