# Spatial Attention Supervision for Defect Localization: Exploiting Ground-Truth Masks as Training Signal in Diffusion-Augmented Defect Detection

arXiv: https://arxiv.org/abs/2609.06232 · HTML: https://mlsanity.com/paper/2609.06232/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 25 of its 34 references that are in the map.
- First posted on arXiv 5 Sep 2026.

## Builds on (most cited first, 25 of 25)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Diffusion Models Beat GANs on Image Synthesis](https://mlsanity.com/paper/2105.05233/) (2021)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://mlsanity.com/paper/1905.11946/) (2019)
- [MobileNetV2: Inverted Residuals and Linear Bottlenecks](https://mlsanity.com/paper/1801.04381/) (2018)
- [Improved Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2102.09672/) (2021)
- [A ConvNet for the 2020s](https://mlsanity.com/paper/2201.03545/) (2022)
- [Learning Deep Features for Discriminative Localization](https://mlsanity.com/paper/1512.04150/) (2015)
- [Visualizing and Understanding Convolutional Networks](https://mlsanity.com/paper/1311.2901/) (2013)
- [Swin Transformer V2: Scaling Up Capacity and Resolution](https://mlsanity.com/paper/2111.09883/) (2021)
- [Towards Total Recall in Industrial Anomaly Detection](https://mlsanity.com/paper/2106.08265/) (2021)
- [Captum: A unified and generic model interpretability library for PyTorch](https://mlsanity.com/paper/2009.07896/) (2020)
- [Right for the Right Reasons: Training Differentiable Models by Constraining their Explanations](https://mlsanity.com/paper/1703.03717/) (2017)
- [FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flows](https://mlsanity.com/paper/2111.07677/) (2021)
- [Tell Me Where to Look: Guided Attention Inference Network](https://mlsanity.com/paper/1802.10171/) (2018)
- [EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies](https://mlsanity.com/paper/2303.14535/) (2023)
- [AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model](https://mlsanity.com/paper/2312.05767/) (2023)
- [Attention Branch Network: Learning of Attention Mechanism for Visual Explanation](https://mlsanity.com/paper/1812.10025/) (2018)
- [Self-produced Guidance for Weakly-supervised Object Localization](https://mlsanity.com/paper/1807.08902/) (2018)
- [The MVTec AD 2 Dataset: Advanced Scenarios for Unsupervised Anomaly Detection](https://mlsanity.com/paper/2503.21622/) (2025)
- [DefectFill: Realistic Defect Generation with Inpainting Diffusion Model for Visual Inspection](https://mlsanity.com/paper/2503.13985/) (2025)
- [Enhancing Glass Defect Detection with Diffusion Models: Addressing Imbalanced Datasets in Manufacturing Quality Control](https://mlsanity.com/paper/2505.03134/) (2025)
- [Region-Aware CAM: High-Resolution Weakly-Supervised Defect Segmentation via Salient Region Perception](https://mlsanity.com/paper/2506.22866/) (2025)
- [Machine Learning in Industrial Quality Control of Glass Bottle Prints](https://mlsanity.com/paper/2409.20132/) (2024)

## Abstract (arXiv)

Ground-truth defect masks in industrial inspection datasets are typically reserved for evaluation. This paper repurposes them as spatial supervision signals during training of classification networks, teaching a model not just what to predict but where to look. The method adds an activation-based attention alignment loss that steers convolutional feature maps toward defect regions, in a mixed-supervision formulation that also accommodates samples without masks, such as diffusion-generated images. Combined with DDPM augmentation, synthetic images contribute quantity while masks contribute spatial precision. We evaluate 85 models (four CNN backbones under a 2x2 data/training factorial over five seeds, plus a Swin-V2-T transformer baseline) on the MVTec-AD bottle benchmark, with localization measured on held-out defect images excluded from classifier gradient updates. Main findings: (1) attention-guided training improves activation-based localization (Pixel-AUROC) by +18.0% for EfficientNetB0 with augmentation (p=0.005, Cohen's d=2.6) and +18.7% for ResNet50 (p=0.008), significant in four of eight CNN settings (uncorrected for multiple comparisons) with no significant change in classification; (2) for EfficientNetB0 a data x training-mode interaction is significant (p=0.002), consistent with a super-additive effect (+13.6% combined vs +1.6% summed individual effects); (3) architectures with weaker spatial representations benefit most, whereas ConvNeXt-T shows no effect, apparently because its depthwise-convolution activations yield spatially uninformative channel-mean maps; (4) unsupervised PatchCore remains the strongest localizer (Pixel-AUROC=0.983), contextualizing the supervised gains. These results show that existing evaluation masks can act as practical training signals that measurably and reproducibly improve where defect classifiers attend.