Unsupervised UNet for Fabric Defect Detection

Currently, neural network based defect detection systems usually need to collect a large number of defect samples for training, and it takes a lot of manpower to mark labels and clean the subsequent data. This is a time-consuming process, and it makes the whole system less effective. In this paper, a neural network based method for fabric surface defect detection is proposed. By training positive clean samples, it can learn through neural network without collecting negative defective samples, which greatly shortens the landing time of whole system. Our proposed system can achieve 99% detection accuracy.