Contamination detection: Datasets & Models

Manual contamination inspection in manufacturing and QC depends on human eyes at the end of a line, which fatigue, miss small particles, and produce inconsistent judgments between shifts and inspectors. Fine debris and residue are easy to overlook at line speed, and a single missed contaminant can trigger recalls, scrap, or customer rejections. Automating this with a segmentation model gives you consistent, repeatable detection that flags contaminants in real time and logs exactly where and how large each one is. That turns subjective visual checks into auditable data you can act on immediately.

A trained RF-DETR instance segmentation model catches contamination earlier and more consistently than periodic manual sampling, reducing scrap and downstream defects while creating a traceable record of every flagged unit. It runs on the cameras and edge hardware you likely already have on the line, and scales across multiple stations, product lines, or facilities without adding headcount. As you gather more images, you can retrain and improve accuracy for your specific products and contaminant types.

1. Upload an image: Drop any photo into the widget above to try contamination detection for free, right on this page.
2. Preview instant results: The model detects metal shaving, dust particle, hair, oil residue, plastic fragment 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 (instance segmentation) 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.