Skelter Labs’ recent innovation in vision technology is the deep learning based Defect Detection Engine, an automated visual inspection tool for the manufacturing industry to improve product quality and operational efficiency. The Defect Detection Engine identifies flawed products with high accuracy, matching human level or better.
- By adopting active normalization technology for complicated shapes, the Defect Detection Engine model can be successfully trained even with a small number of defected product data
- Using image normalization & pre-processing techniques for data optimization, the AI inference engine can effectively deliver higher performance compared to other known solutions
- Proven performance track record of over 99% defect detection accuracy for automotive supplier parts with a metal surface
- A wide range of various defects such as scratches, cracks, nicks, and dents are identified and categorized in real-time
- Applying various deep learning models for image classification, localization, and segmentation that deliver a unique deep learning based defect detection model, specifically strong in micro-defects.