arXiv 2502.198482025-07-22cs.CV

One-for-More：用于异常检测的持续扩散模型

One-for-More: Continual Diffusion Model for Anomaly Detection

East China Normal University（东华师范大学）；Shanghai Innovation Institute（上海创新研究院）；Xiamen University（厦门大学）；Shanghai Jiao Tong University（上海交通大学）；Tencent YouTu Lab（腾讯YouTu实验室）；Tencent WeChatPay Lab33（腾讯微信支付实验室）；CATL（宁德时代）；Key Laboratory of Multimedia Trusted Perception and Efficient Computing Ministry of Education of China（教育部多媒体可信感知与高效计算重点实验室）

Xiaofan Li, Xin Tan, Zhuo Chen, Zhizhong Zhang, Ruixin Zhang, Rizen Guo, Guannan Jiang, Yulong Chen, Yanyun Qu, Lizhuang Ma, Yuan Xie

备注：: Accepted by CVPR2025

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文针对扩散模型在持续异常检测中的保真幻觉与灾难性遗忘问题，提出基于梯度投影的持续扩散模型，并通过迭代奇异值分解降低内存开销，同时引入异常掩码网络增强条件机制，在 MVTec 和 VisA 上取得领先性能。

摘要：

随着生成模型的兴起，人们越来越关注在生成框架内统一所有任务。异常检测方法也属于这一范畴，并利用扩散模型在给定任意异常图像时生成或重建正常样本。然而，我们的研究发现，扩散模型存在严重的“保真幻觉”和“灾难性遗忘”问题，无法应对不可预测的模式增量。为了缓解上述问题，我们提出了一种持续扩散模型，该模型使用梯度投影来实现稳定的持续学习。梯度投影通过将梯度修改为保护已学知识的方向，对模型更新施加正则化。但作为一把双刃剑，它也需要由马尔可夫过程带来的巨大内存开销。因此，我们提出了一种基于线性表示传递性质的迭代奇异值分解方法，该方法消耗极少内存，并且几乎不造成性能损失。最后，考虑到扩散模型对正常图像“过拟合”的风险，我们提出了一种异常掩码网络来增强扩散模型的条件机制。对于持续异常检测，我们的方法在 MVTec 和 VisA 的 17/18 个设置中取得了第一名。代码可在 https://github.com/FuNz-0/One-for-More 获取。

英文摘要：

With the rise of generative models, there is a growing interest in unifying all tasks within a generative framework. Anomaly detection methods also fall into this scope and utilize diffusion models to generate or reconstruct normal samples when given arbitrary anomaly images. However, our study found that the diffusion model suffers from severe ``faithfulness hallucination'' and ``catastrophic forgetting'', which can't meet the unpredictable pattern increments. To mitigate the above problems, we propose a continual diffusion model that uses gradient projection to achieve stable continual learning. Gradient projection deploys a regularization on the model updating by modifying the gradient towards the direction protecting the learned knowledge. But as a double-edged sword, it also requires huge memory costs brought by the Markov process. Hence, we propose an iterative singular value decomposition method based on the transitive property of linear representation, which consumes tiny memory and incurs almost no performance loss. Finally, considering the risk of ``over-fitting'' to normal images of the diffusion model, we propose an anomaly-masked network to enhance the condition mechanism of the diffusion model. For continual anomaly detection, ours achieves first place in 17/18 settings on MVTec and VisA. Code is available at https://github.com/FuNz-0/One-for-More

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。