AI-powered quality control at Royal Greenland

The Challenge

A need for intelligent quality inspection

Quality control is the process of validating whether a product or service meets the expected delivery standards. Today, many industries do quality control by hand, and that’s the case for Royal Greenland’s frozen prawn product lineup.

While highly methodical, their manual quality process comes with significant challenges:

It’s tiring: Manual inspections are repetitive and require countless micro-decisions for every analyzed sample.
It’s inconsistent: Lightning varies, rooms differ, and human perception is not uniform.
It’s prone to noise: Biases, fatigue, stress, and emotions all affect judgment.

The outcome is introducing variability for a process that requires consistency and reliability. The challenge was clear:

How could Trifork and Royal Greenland build a system that standardizes the quality grading process and minimizes the variability of outcomes?

The Solution

An advanced quality control system merging AI, hardware and SAP

Solving the challenge needed to build a system that captures prawn samples as pictures in controlled environmental conditions, analyzes them under the same evaluation criteria, and presents the results in an easy and friendly way. To do it, we merged three different factors:

Vision AI — We trained advanced AI image recognition models with deep learning capabilities to analyze, measure, and grade prawn samples.
High-End Hardware — We built an enclosed cabinet that isolates prawn samples from external factors, creating stable evaluation conditions. The cabinet hosts high-end hardware — camera, scale, lights — to record samples as images; all powered by an industrial AI computer from Lenovo.
Software — We developed a web platform powered by SAP Fiori that presents AI results to lab technicians, and we use SAP BTP to host all results, working as the bridge for the web platform and all hardware.

All combined results in an end-to-end quality control system that helps lab technicians remove guesswork and makes analysis results transparent, uniform, and trusted.

Our approach

A process-first approach with end-users in focus

User-Centric Value — Running research activities at Royal Greenland’s Cuxhaven and Aalborg plants, to validate needs and perceived challenges, and better understand the quality control process and its end-users.
Controlled Delivery — To develop a product that doesn’t exist in the market and relies on R&D, we adopted a controlled Proof In Production delivery model — defining a lean scope to validate the feasibility and business value of Vision AI for Royal Greenland’s inspection processes.
Hands-On Expertise — To cover a wide range of challenges, our interdisciplinary team — including project and product management, solution architecture, experience design, and AI engineering — worked closely across business and technical domains on-site in Cuxhaven, Germany.

The Outcome

Smart quality control at the press of a button

While the product is in its early stages, preliminary outcomes are highly promising.

Pushing the manual inspection limits — bringing intelligence directly to lab technicians’ fingertips.
- Standardized color grading of prawns with data-driven precision.
- Classify the morphology of hundreds of prawns and count them in a matter of seconds, not minutes.
- Count hundreds of prawns significantly faster than manual checks, and yet with the same accuracy.
Traceability of evaluations — Thanks to SAP BTP, all pictures captured through the cabinet and their analysis results are saved in the cloud. This is a game-changer for Royal Greenland, enabling the review of past evaluations for the first time and building a database to help train future AI models.
Strong Corporate Social Responsibility (CSR) benefit — By empowering Greenlandic workers with better digital tools, the solution reinforces Greenland’s role in the global seafood supply chain.

Why Trifork

Royal Greenland choose Trifork for our:

End-to-end expertise across AI and SAP
Strong Proof of Concept delivery model
SAP-native UX for high adoption
Built for scale and reuse
Trusted AI partnerships with Lenovo and NVIDIA

AI-Driven Image Recognition

Custom-trained Vision AI model for detecting visual defects in seafood batches.

End-to-End SAP Integration

AI results presented through a SAP-native UX tailored for lab technicians.

Built on SAP BTP

Application conceptualized as a reusable product running on SAP Business Technology Platform.

Curious on how we could help you transform your quality control operations?

If you prefer, you can also contact us on info@trifork.com