mlsanityData to 7 Oct 2026

Statistical Non-linear Reconstruction Loss for Image Anomaly Detection

arXiv 2607.12866 ↗First posted 14 Jul 2026cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Reconstruction-based methods are a cornerstone of unsupervised image anomaly detection, but they remain vulnerable to 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.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 22 of its 29 references that are in the map. First posted on arXiv 14 Jul 2026.

Builds on

22 of 29 references.

References · 22

  1. 2021Towards Total Recall in Industrial Anomaly Detectioncited by 772
  2. 2020PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localizationcited by 591
  3. 2018Real-world Anomaly Detection in Surveillance Videoscited by 570
  4. 2019Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detectioncited by 560
  5. 2021CutPaste: Self-Supervised Learning for Anomaly Detection and Localizationcited by 529
  6. 2021DRAEM -- A discriminatively trained reconstruction embedding for surface anomaly detectioncited by 510
  7. 2018GANomaly: Semi-Supervised Anomaly Detection via Adversarial Trainingcited by 409
  8. 2022SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentationcited by 397
  9. 2019Uninformed Students: Student-Teacher Anomaly Detection with Discriminative Latent Embeddingscited by 386
  10. 2020Multiresolution Knowledge Distillation for Anomaly Detectioncited by 285
  11. 2022A Unified Model for Multi-class Anomaly Detectioncited by 267
  12. 2018Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoderscited by 263
10 more
  1. 2020Patch SVDD: Patch-level SVDD for Anomaly Detection and Segmentationcited by 212
  2. 2019OCGAN: One-class Novelty Detection Using GANs with Constrained Latent Representationscited by 199
  3. 2023Deep Industrial Image Anomaly Detection: A Surveycited by 147
  4. 2020Deep Learning for Medical Anomaly Detection -- A Surveycited by 127
  5. 2013Generalized Denoising Auto-Encoders as Generative Modelscited by 113
  6. 2024MambaAD: Exploring State Space Models for Multi-class Unsupervised Anomaly Detectioncited by 113
  7. 2023Hierarchical Vector Quantized Transformer for Multi-class Unsupervised Anomaly Detectioncited by 66
  8. 2022ADTR: Anomaly Detection Transformer with Feature Reconstructioncited by 31
  9. 2025Learning to Detect Multi-class Anomalies with Just One Normal Image Promptcited by 18
  10. 2024Salvaging the Overlooked: Leveraging Class-Aware Contrastive Learning for Multi-Class Anomaly Detectioncited by 15

Built on by

Not cited yet.