PADFormer: Pose-agnostic Anomaly Detection from Sparse View Images
This work provides an efficient image-space solution for anomaly detection under arbitrary viewpoints, benefiting industrial inspection and related domains without the need for 3D reconstruction.
PADFormer tackles pose-agnostic anomaly detection by using a Vision Transformer to reconstruct anomaly-free images from sparse views, avoiding costly 3D reconstruction. It achieves state-of-the-art results on the PAD benchmark and comparable performance on few-shot anomaly detection tasks.
Pose-agnostic Anomaly Detection (PAD) remains challenging as anomalies can appear under arbitrary viewpoints, requiring methods to handle significant pose variations. Existing approaches rely on complex 3D reconstruction, which are computationally expensive and require extensive multi-view data. We propose PADFormer, a novel image-space approach that leverages Vision Transformer (ViT) to directly reconstruct anomaly-free versions of query images while preserving pose information. Our key insight is to adapt cross-view masked reconstruction for anomaly detection through training exclusively on normal data, combined with dynamic patch selection and spatial alignment mechanisms that enable effective learning from sparse reference views under significant pose variations. During inference, we perform multiple forward passes with different masking patterns to generate an ensemble of anomaly-free reconstructions, ensuring comprehensive coverage of the query image. Anomalies are detected by comparing these reconstructions with the query image. PADFormer achieves state-of-the-art results on the PAD benchmark while maintaining comparable performance on classic few-shot anomaly detection (FSAD) tasks, demonstrating superior efficiency and generalization without requiring 3D reconstruction.