Glass defect detection: Datasets & Models

Manual inspection of glass is slow, inconsistent, and hard to scale across high-throughput production lines, and human inspectors fatigue when scanning for subtle scratches or hairline cracks under bright lighting. In manufacturing and QC environments, a missed defect can lead to costly downstream rework, customer returns, or safety issues with tempered and structural glass. Automating glass defect detection with a vision model gives you consistent, repeatable inspection at line speed and creates an audit trail of every flagged unit. It also lets you catch defect trends early and correct upstream process problems before they generate large batches of scrap.

An object detection model like RF-DETR pinpoints and classifies each defect, so you can act on defect type and location rather than a simple pass/fail signal. It integrates with existing inline cameras and PLC-driven reject mechanisms, and scales across multiple lines and facilities without adding inspection headcount. Continuous data capture also builds a labeled dataset that improves model accuracy over time and supports root-cause analysis of recurring defects.

1. Upload an image: Drop any photo into the widget above to try glass defect detection for free, right on this page.
2. Preview instant results: The model detects crack, chip, scratch, bubble, glass panel 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.