Surface Defect Detection Methods for Industrial Products: A Review

The comprehensive intelligent development of the manufacturing industry puts forward new requirements for quality inspection industrial products. This paper summarizes current research status machine learning methods in surface defect detection, a key part First, according to use features, application traditional vision detection product is summarized from three aspects: texture color and shape features. Secondly, based on deep technology recent years discussed supervised method, unsupervised weak method. Then, common problems their solutions are systematically summarized; include real-time problem, small sample target unbalanced problem. Lastly, commonly used datasets defects more comprehensively summarized, latest MVTec AD dataset compared, so as provide some reference further technology.