Learning to Detect Multi-class Anomalies with Just One Normal Image Prompt
This work addresses a critical issue in industrial anomaly detection by improving generalization and segmentation accuracy, though it is incremental as it builds on existing reconstruction-based approaches.
The paper tackles the problem of multi-class anomaly detection using reconstruction networks, which often fail due to perfect reconstruction of anomalies and low-resolution segmentation. The proposed OneNIP method reconstructs normal features and restores anomalies with just one normal image prompt, outperforming previous methods on benchmarks like MVTec, BTAD, and VisA.
Unsupervised reconstruction networks using self-attention transformers have achieved state-of-the-art performance for multi-class (unified) anomaly detection with a single model. However, these self-attention reconstruction models primarily operate on target features, which may result in perfect reconstruction for both normal and anomaly features due to high consistency with context, leading to failure in detecting anomalies. Additionally, these models often produce inaccurate anomaly segmentation due to performing reconstruction in a low spatial resolution latent space. To enable reconstruction models enjoying high efficiency while enhancing their generalization for unified anomaly detection, we propose a simple yet effective method that reconstructs normal features and restores anomaly features with just One Normal Image Prompt (OneNIP). In contrast to previous work, OneNIP allows for the first time to reconstruct or restore anomalies with just one normal image prompt, effectively boosting unified anomaly detection performance. Furthermore, we propose a supervised refiner that regresses reconstruction errors by using both real normal and synthesized anomalous images, which significantly improves pixel-level anomaly segmentation. OneNIP outperforms previous methods on three industry anomaly detection benchmarks: MVTec, BTAD, and VisA. The code and pre-trained models are available at https://github.com/gaobb/OneNIP.