Leather Defect Detection

Clients

Techstack

Purpose

Spotting surface flaws in leather by eye is slow and easy to get wrong, and a missed defect can ruin a finished product downstream. The university wanted an affordable way to catch flawed material early, before it costs anyone anything.

Description

We built a compact vision prototype that inspects leather as it's viewed, locating and labeling surface defects in real time with a custom-trained deep-learning model running on low-cost hardware. An on-screen interface shows the live inspection with detections marked, and switchable viewing modes make it easy to compare flagged and clean material. Packaged to install and run on inexpensive gear, it shows how we bring real-time quality-control vision within a small budget.