Codon Consulting | Project | Computer vision for hazard detection in scrap metal

As steelmaking moves to electric arc furnaces, recycled scrap becomes central to production. For a steel producer, sealed pressure vessels hidden in incoming scrap could cause furnace explosions, and manual inspection of cluttered, high-volume material did not scale.

‍

We built an edge computer vision system that detects hazardous objects in cluttered scrap under live operating conditions. It also classifies scrap, estimates how full containers are and predicts chemical composition and quality, and it keeps improving through an active learning loop that learns from operator feedback.

[data-wf-bgvideo-fallback-img] { display: none; } @media (prefers-reduced-motion: reduce) { [data-wf-bgvideo-fallback-img] { position: absolute; z-index: -100; display: inline-block; height: 100%; width: 100%; object-fit: cover; } }

As steelmaking moves to electric arc furnaces, recycled scrap becomes central to production. For a steel producer, sealed pressure vessels hidden in incoming scrap could cause furnace explosions, and manual inspection of cluttered, high-volume material did not scale.

‍

We built an edge computer vision system that detects hazardous objects in cluttered scrap under live operating conditions. It also classifies scrap, estimates how full containers are and predicts chemical composition and quality, and it keeps improving through an active learning loop that learns from operator feedback.

Value for the client:

Hazardous objects detected before material reaches the furnace
Automated classification of scrap metal
Container fill levels and material quality estimated automatically
A model that gets better with operator feedback, rather than degrading

No items found.

Case Studies

Discover how we solve complex challenges with AI.

All Projects

Transforming pest control with Al-driven IoT sensor analysis

A leading pest control company faced growing challenges in managing vast volumes of sensor data generated around the clock.

Simplifying document digitization with AI-based OCR system

One of Sweden’s three nuclear power plants expressed a need to digitize a vast amount of classified documents.

Property Intelligence Solution Based on Multimodal Sensor Data

A leading Stockholm property owner is advancing prop tech with sensor-driven data. We partnered with their teams to deliver AI models for smarter space usage and optimized district heating.