MagicQC
Every piece measured to each brand’s tolerance table, logged to the operator.
- Client
- Confidential
- Year
- 2025
- Category
- Computer Vision
- Role
- Design + Build
- Timeline
- Ongoing
The problem
Garment QC ran on a tape measure and a stack of paper spec sheets. adidas, Zara, Puma and Reebok each bring their own list of points of measure and their own tolerance per size, and operators reconciled all of it by hand. Mistakes shipped, and the QC manager only found out when a customer complained.
What we built
A desktop capture app paired with a web dashboard. The operator picks brand, article type, style and size; the station loads that brand’s points of measure with the tolerance for that size. Each measurement is captured, compared to spec, and marked pass or fail in centimetres at the point of measure, with garment colour and size driving the tolerance set. “Next Piece” and “Next Article” keep the line moving, and every result syncs to the web app with the operator and table attached.
The result
Operators stopped looking up specs, because the station already has them. Pass and fail are logged per operator and per table, and the QC manager watches results arrive on the web app instead of collecting clipboards at the end of a shift. In production, and presented at MyKarachi, the KCCI international expo.
- garments measured on a live line
- 500+ / shiftgarments measured on a live line
- each with its own POM list and tolerances
- 8+ brandseach with its own POM list and tolerances
- in centimetres, at the point of measure
- pass / failin centimetres, at the point of measure
Built with
- Next.js
- TypeScript
- Python
- OpenCV
- PostgreSQL
- Electron
- AWS EC2
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.