arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.09533cs.CV
收藏

AnomalyFactory:将异常生成视为无监督异常定位

AnomalyFactory: Regard Anomaly Generation as Unsupervised Anomaly Localization

Ricoh Software Research Center (Beijing) Co., Ltd.(理光软件研究中心(北京)有限公司)

Ying Zhao

AI总结:

针对现有异常生成方法需多模型适配不同数据与类别、扩展性差的问题,提出可扩展框架AnomalyFactory,用同一网络架构统一无监督异常生成与定位,在5个数据集上验证了其生成能力与可扩展性的优势。

摘要:

异常生成方法的最新进展缓解了数据不足对异常定位任务的影响。这类方法虽然有效,但大多数都需要在不同数据集上训练多个大型生成模型,还要为不同类别搭建繁琐的异常预测模型。为解决这些局限,我们提出了一种新型可扩展框架AnomalyFactory,它用同一网络架构统一了无监督异常生成与异常定位任务。该框架首先通过BootGenerator(引导生成器),在学习得到的热图引导下,融合目标边缘图的结构与参考彩色图像的外观;随后由FlareGenerator(耀斑生成器)接收BootGenerator的监督信号,调整热图以指示生成图像中的异常位置;最后,它可轻松将同一网络架构转换为BlazeDetector(烈焰检测器),通过将FlareGenerator生成的异常图像转换为正常图像,借助学习到的热图定位异常像素。通过调整目标边缘图并将其与各类参考图像结合,AnomalyFactory能够生成跨域的真实且多样化的样本。我们在MVTecAD、VisA、MVTecLOCO、MADSim和RealIAD共5个数据集上开展了综合实验,结果表明该方法在生成能力和可扩展性上优于同类方法。

英文摘要:

Recent advances in anomaly generation approaches alleviate the effect of data insufficiency on task of anomaly localization. While effective, most of them learn multiple large generative models on different datasets and cumbersome anomaly prediction models for different classes. To address the limitations, we propose a novel scalable framework, named AnomalyFactory, that unifies unsupervised anomaly generation and localization with same network architecture. It starts with a BootGenerator that combines structure of a target edge map and appearance of a reference color image with the guidance of a learned heatmap. Then, it proceeds with a FlareGenerator that receives supervision signals from the BootGenerator and reforms the heatmap to indicate anomaly locations in the generated image. Finally, it easily transforms the same network architecture to a BlazeDetector that localizes anomaly pixels with the learned heatmap by converting the anomaly images generated by the FlareGenerator to normal images. By manipulating the target edge maps and combining them with various reference images, AnomalyFactory generates authentic and diversity samples cross domains. Comprehensive experiments carried on 5 datasets, including MVTecAD, VisA, MVTecLOCO, MADSim and RealIAD, demonstrate that our approach is superior to competitors in generation capability and scalability.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑