Ai-powered imaging system for objective quality assessment of meat in gravy and jelly

Overview

This technology is an integrated inspection system designed for food manufacturers producing meat products in gravy or jelly. It combines portable controlled illumination with RGB and multispectral cameras to capture standardised images with minimal operator input. Machine-learning segmentation distinguishes meat pieces, gravy or jelly, and packaging, enabling automated analysis of piece count, size distribution, shape, calibrated meat colour, colour uniformity, and meat-to-background ratio. The system issues product-specific pass/review/fail alerts and generates annotated images, colour maps, measurements, and batch metadata for traceability and trend analysis.

Technical specifications

The system builds on existing expertise in food imaging, spectroscopy, and artificial intelligence. It uses machine-learning models for segmentation and quality prediction. RGB vision measures morphology and meat-to-background ratio; calibrated colour mapping quantifies product colour; multispectral data improve separation where visible contrast is limited. The system is designed to integrate at-line or post-fill positions in production lines and can run on edge computing for real-time decisions. Outputs include a demonstrator, operating procedure, and traceable reporting for quality assurance.

Technology readiness level

Currently at concept stage with validated component methods. The team has developed colour-based machine-learning models for meat marbling prediction, multispectral-ML methods for avocado-oil quality, and NIR-ML methods for food assessment through packaging. Future validation includes collecting images across batches, training AI models, integrating hardware into a prototype, and conducting factory trials to assess accuracy, repeatability, speed, usability, cleanability, and agreement with quality-assurance decisions. This is an early-stage technology with a clear roadmap to scale-up.