NU Smart Farms

Automated screening for pig pathologies

Background

The problem

EFSA has introduced visual inspection only of pig carcasses at abattoirs to ensure quality and safety of meat products.

This puts a strain on human resources and involves subjectivity. Would automated detection be a solution?

Research project

The proof of principle

Develop an automated system based on visual image analysis to screen pig offal at the abattoir. The system would:

  • detect public health hazards and subclinical diseases
  • provide automated capture of pathologies

The computer vision pipeline should:

  • identify offal region
  • localise individual organs
  • detect organs with specific conditions

To ensure accuracy, agreement between machine vision and meat inspection experts would need to be quantified.