Visual Inspection System for Anomaly Detection on KTL Coatings Using Variational Autoencoders

Abstract

Electric cathode metal coating (KTL) is a popular choice for surface protection of metal components in the automotive industry. Due to the complex 3D shape of the parts and the glossy black color of the coating, machine vision inspection is very sensitive to variabilities among parts and to the variabilities in their positioning during the image acquisition. In this paper a variational autoencoder model for anomaly detection is presented to make further image processing more immune to variability and to detect coating defects more reliably.