本项目实现了一种基于 VAE-CycleGAN 的图像重建无监督缺陷检测算法。该算法结合了变分自编码器 (VAE) 和 CycleGAN 的优势，无需标注数据即可检测图像中的缺陷/异常。This project implements an unsupervised defect detection algorithm for image reconstruction based on VAE-CycleGAN. This algorithm combines the advantages of variational autoencoders (VAE) and CycleGAN to detect defects in images without any supervision.