用非线性损失抑制异常重建，提升工业质检中的异常定位精度
Statistical Non-linear Reconstruction Loss for Image Anomaly Detection
- 引入基于Sigmoid的非线性重建损失，压制高幅值特征以防止异常干扰训练
- 在MVTec-AD上达到99.0%图像级和97.3%像素级AUROC，VisA上达95.3%和99.0%
- 通过统计校准自动选择抑制强度，适合需要精准异常定位的工业检测场景
基于重建的无监督图像异常检测方法虽为核心，但仍易受异常泄漏影响，即标准均方误差（MSE）损失会促使模型忠实重建异常模式。本文提出一种非线性重建损失，采用Sigmoid型压缩函数抑制高幅值特征，防止异常主导优化过程，同时保留对正常模式的敏感性。此外，引入统计校准方案，从正常特征分布的置信区间（CI）中选取缩放因子 $k$，实现数据驱动的抑制强度控制。所提方法在性能上优于或媲美现有先进方法，在MVTec-AD上达到99.0%图像级AUROC与97.3%像素级AUROC；在VisA数据集上达到95.3%图像级与99.0%像素级AUROC。结果表明，非线性梯度抑制是缓解异常泄漏、提升统一工业检测中异常定位能力的有效机制。代码已开源：https://github.com/mintii13/Statistical-Non-linear-Reconstruction-Loss.git。
原文摘要 · Abstract (English)
Reconstruction-based methods are a cornerstone of unsupervised image anomaly detection, but they remain vulnerable to \emph{outlier leakage}, where standard mean squared error (MSE) loss drives the model to faithfully reconstruct anomalous patterns. We propose a Non-linear Reconstruction Loss that applies a sigmoid-based squashing function to suppress high-magnitude features, preventing outliers from dominating optimization while preserving sensitivity to normal patterns. In addition, we introduce a statistical calibration scheme that selects the scaling factor $k$ from the confidence interval (CI) of the normal feature distribution, enabling data-driven control of the suppression strength. Our approach achieves competitive or superior anomaly detection performance compared to state-of-the-art methods, reaching 99.0\% Image-AUROC and 97.3\% Pixel-AUROC on MVTec-AD, and 95.3\% Image-AUROC and 99.0\% Pixel-AUROC on VisA. These results indicate that non-linear gradient suppression is an effective mechanism for mitigating outlier leakage and improving anomaly localization in unified industrial inspection settings. The implementation is available at https://github.com/mintii13/Statistical-Non-linear-Reconstruction-Loss.git.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。