Ai-powered automated visual quality assessment for meat-in-gravy and meat-in-jelly products

Overview

UBI Meat's AI computer-vision system automatically analyzes images of meat-in-gravy and meat-in-jelly products to deliver objective, repeatable quality measurements. The system quantifies meat-piece size, shape, and color, as well as background (gravy/jelly) color and the meat-to-background ratio. By segmenting images and performing object-level analysis, it generates reproducible measurements, statistical distributions, and deviation flags against product specifications—replacing subjective, manual visual assessment with consistent, documentable results.

The technology is a direct extension of UBI Meat's proven decision-grading platform, which already uses artificial intelligence and machine learning to deliver objective meat quality and composition assessment in real food-processing environments. The same core capabilities—image capture, segmentation, object measurement, classification, and specification-based deviation detection—are adapted to assess meat pieces in gravy/jelly. This gives processors a practical path from rapid at-line or near-line validation toward inline inspection.

Key benefits:

• Objective and repeatable quality measurements replace subjective grading
• Consistent, recordable documentation of quality standards
• Automated deviation alerts against product specifications
• Supports routine QA workflows with minimal operator input

Technical specifications

How it works:

• Image capture using rapid at-line/near-line imaging, with a defined pathway to inline inspection
• AI/ML segmentation models separate meat pieces from the gravy/jelly background and quantify relevant color attributes
• Object-level analysis measures meat-piece size, shape, color, background color, and meat-to-background ratio
• Results are compiled into repeatable measurements, distributions, and deviation flags against product specifications
• Automated reporting and deviation alerts support routine QA workflows

Key features:

• Models trained on representative images spanning batches, plants, operators, and normal process variation
• Standardized imaging conditions defined for consistency
• Validation against reference measurements and blinded samples to quantify accuracy, repeatability, false flags, and processing time
• Architecture supports expansion to additional SKUs and broader deployment

Technology readiness level

The core computer-vision and segmentation technology is already commercialized by UBI Meat for objective meat quality and composition assessment in real food-processing environments. The application to meat-in-gravy/jelly products is at the proof-of-concept stage, with a structured validation roadmap: aligning on product specifications and sampling points, collecting representative images across process variation, training and adapting segmentation and measurement models, validating against reference and blinded samples, and demonstrating a routine QA workflow with automated reporting and deviation alerts. The deployment path runs from at-line/near-line validation toward inline inspection, with broader SKU expansion planned after initial demonstration.