Case study · Team experience · Industry
Quality control with computer vision on a production line
The problem
On a production line, checking by eye that every box leaves without defects, and separating the different types, depends on one person paying attention for hours.
The solution
A camera over the line that detects defects and tells box types apart in real time, alerting operators so they can act on the spot.
Manual visual checks, with errors slipping through at the end of the shift.
Every box checked as it passes, with an instant alert if something is wrong.
Context
Factories and warehouses move hundreds of boxes an hour. A cardboard defect or the wrong box slipping through ends up as a badly served order or a return.
The challenge
Checking depends on the operators' eyes and attention, and they have other tasks too. Errors are discovered late, once the product has already left.
How we approached it
- An industrial camera and suitable lighting over the key point of the line.
- A vision model trained on real images of the product: dents, tears, bad folds or wrong labels.
- Each box classified by type or format, so references are never mixed.
- An instant alert to operators on screen or with a light signal, or automatic diversion of the box.
- An incident log and per-shift statistics to spot patterns.
Technology
- Computer vision
- Industrial cameras
- On-site processing
- Web dashboard
The result
Fewer defective products reach the customer and operators know instantly what to check, without relying on their eyes all shift long. The same idea works for parts, packaging or any product that passes in front of a camera.
A project built by people on our team. For confidentiality we do not show the client or technical details.
Facing a similar challenge?
Tell us what you need and we'll call you. Free 30-minute first meeting, no strings attached.
Let's talk