Tire-Cord Fabric Defect Detection
Flagged live breaks, gaps and weave irregularities called out as the fabric runs.
- Client
- Confidential
- Year
- 2025
- Category
- Computer Vision
- Role
- Design + Build
- Timeline
- 4–5 weeks
The problem
Tire-cord fabric defects are subtle: a broken cord, a gap, an irregular weave, all easy to miss at line speed. Missing one puts a weak spot into a tyre.
What we built
An AnomalyCLIP model trained on the fabric renders a defect heatmap next to the live feed: normal weave stays cool, anomalies light up. It does not need every defect type labelled in advance. It learns what normal looks like, then flags anything that departs from it.
The result
A second set of eyes on the fabric that never blinks, with a heatmap an operator can act on immediately.
- per-frame anomaly score over the weave
- heatmapper-frame anomaly score over the weave
- learns normal, doesn’t need every defect labelled
- few-shotlearns normal, doesn’t need every defect labelled
- runs on the inspection camera feed
- line-speedruns on the inspection camera feed
Built with
- Python
- AnomalyCLIP
- Deep learning
- OpenCV
- Roboflow
We deliver what we commit.
Tell us what you're trying to build.
We'll come back within 24 hours with honest feedback on scope, timeline and cost, whether or not we turn out to be the right fit.