Industrial defect detection method and system

• Main Authors: SHEN WEIMING, JIANG YUXIN, CAO YUNKANG
• Format: Patent
• Language: Chinese; English
• Published: 11.08.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses an industrial defect detection method and system, belongs to the technical field of industrial image processing, and provides a normal feature reconstruction model based on an image level and feature level mask strategy, which can reconstruct normal features according to abnormal features. In an image level mask strategy, extracting normal and abnormal features from a normal industrial image and an abnormal industrial image constructed based on the normal image, and respectively taking the normal and abnormal features as an input signal and a supervision signal; the difference between the input signal and the supervision signal promotes the model to learn global information so as to guide the reconstruction of the abnormal region. In a feature level mask strategy, after feature extraction is carried out on an abnormal feature map after fusion and compression, part of features of the obtained feature map are masked randomly, and then a missing part is reconstructed so as to emphasize lo