Slag detection: Datasets & Models

Slag carryover during tapping and casting degrades metal quality, damages refractory lining, and forces expensive reprocessing or scrap. The manual alternative relies on operators watching bright, high-temperature pour streams for the moment slag breaks through, which is inconsistent across shifts and dangerous to observe up close. Automating slag detection with computer vision gives a consistent, tireless second set of eyes on every pour, catching carryover the instant it starts and enabling faster slag-stop response. For mining and metals operations, that translates directly into higher yield, better metallurgical consistency, and reduced refractory wear.

Automated slag detection improves pour consistency, reduces contamination-driven rework, and extends refractory life by enabling earlier cutoff. It works with the optical and thermal cameras already mounted on furnaces, ladles, and casting lines, so there is no need to re-engineer the pour floor. Built on RF-DETR, the detection model can scale from a single tapping station to every casting line in the plant while maintaining reliable performance in challenging high-glare, high-heat conditions.

1. Upload an image: Drop any photo into the widget above to try slag detection for free, right on this page.
2. Preview instant results: The model detects slag, molten metal stream, ladle, tundish, pour spout within seconds and returns an annotated image with structured predictions, so you can immediately see how well it works. Adjust the detection classes and re-run until the results fit your scenario.
3. Like what you see?: With a free Roboflow account, you can deploy this workflow with the Serverless Cloud API, customize it in your workspace, or train RF-DETR on your own images for production accuracy.
4. Get started for free: No code or training data required to begin: upload, review the results, and make it part of your stack in a single click.