Real-Time Visual Inspection for Poultry Processing

Computer vision MVP for automated fat segmentation and quality grading of turkey carcasses on a live production line — (2022).

Client: IaLINK — AI solutions for food production & health
Funding: MVP — IaLINK VERIA-CLOUD
Role: Machine Learning Developer

Context

IaLINK is a Chilean technology company specializing in AI-powered inspection systems for manufacturing and food production. Their flagship industrial platform, VERIA-CLOUD, targets high-speed, repetitive production processes where human visual inspection is too slow or inconsistent to meet quality standards — replacing it with real-time computer vision that monitors continuously and flags anomalies automatically.

This project extended VERIA-CLOUD into poultry processing, one of Chile’s largest food production sectors, in collaboration with the country’s leading turkey processor.

Objective

Develop a real-time in-line inspection tool capable of operating directly within an active turkey processing facility — under strict requirements for inspection accuracy, throughput, and speed. As carcasses move along the production line in front of a camera, the system must automatically segment fat regions on each carcass, estimate a fat percentage per unit, and identify the appropriate cutting zones based on fat distribution.

The system operates at production-line speed with no manual intervention, providing per-carcass quality data that enables objective, consistent grading at a scale and pace that human inspectors cannot match.

Real-time fat segmentation on poultry carcasses. The algorithm detects and segments fat regions frame-by-frame as units pass the camera, computing a fat percentage estimate and flagging optimal cutting zones for downstream processing.

Status: MVP — currently in validation phase within a live production environment.