arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.06285cs.CV
收藏

以序列重构进行上下文增强的统一无监督异常检测

Context Enhancement with Reconstruction as Sequence for Unified Unsupervised Anomaly Detection

Tsinghua University(清华大学);Hangzhou Zhuoxi Institute of Brain and Intelligence(杭州浙西脑与智能研究院);Xinjiang University(新疆大学)

Hui-Yue Yang, Hui Chen, Lihao Liu, Zijia Lin, Kai Chen, Liejun Wang, Jungong Han, Guiguang Ding

AI总结:

针对统一无监督异常检测中特征重构上下文感知不足的问题,本文提出 RAS,并通过 RASFormer 模块建模空间关系与序列依赖,显著提升检测性能。

摘要:

无监督异常检测(AD)旨在仅使用正常样本训练鲁棒的检测模型,同时能够很好地泛化到未见过的异常。近期研究聚焦于一种统一的无监督 AD 设置,即仅为所有类别训练一个模型,也就是 n-class-one-model(n 类一模型)范式。基于特征重构的方法在该场景下取得了最先进的性能。然而,现有方法往往缺乏足够的上下文感知能力,从而损害重构质量。为解决这一问题,本文提出一种新颖的 Reconstruction as Sequence(RAS,序列即重构)方法,从序列建模的角度增强特征重构过程中的上下文对应关系。具体而言,基于 transformer 技术,我们在 RAS 中集成了专门的 RASFormer 模块。该模块能够捕捉不同图像区域之间的空间关系,并增强整个重构过程中的序列依赖。通过引入 RASFormer 模块,我们的 RAS 方法获得了更强的上下文感知能力,从而带来显著性能提升。实验结果表明,RAS 明显优于竞争方法,充分证明了该方法的有效性和优越性。代码可在 https://github.com/Nothingtolose9979/RAS 获取。

英文摘要:

Unsupervised anomaly detection (AD) aims to train robust detection models using only normal samples, while can generalize well to unseen anomalies. Recent research focuses on a unified unsupervised AD setting in which only one model is trained for all classes, i.e., n-class-one-model paradigm. Feature-reconstruction-based methods achieve state-of-the-art performance in this scenario. However, existing methods often suffer from a lack of sufficient contextual awareness, thereby compromising the quality of the reconstruction. To address this issue, we introduce a novel Reconstruction as Sequence (RAS) method, which enhances the contextual correspondence during feature reconstruction from a sequence modeling perspective. In particular, based on the transformer technique, we integrate a specialized RASFormer block into RAS. This block enables the capture of spatial relationships among different image regions and enhances sequential dependencies throughout the reconstruction process. By incorporating the RASFormer block, our RAS method achieves superior contextual awareness capabilities, leading to remarkable performance. Experimental results show that our RAS significantly outperforms competing methods, well demonstrating the effectiveness and superiority of our method. Our code is available at https://github.com/Nothingtolose9979/RAS.

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。

↑