FeatDAE: Introducing Features With Denoising Autoencoder for Anomaly Detection
Zheyuan Zhou, J. C. Wang, Zian Yu, Zili Wang, Xiaojian Liu, Lemiao Qiu, Shuyou Zhang · IEEE Transactions on Instrumentation and Measurement · 2025
Anomaly detection is a critical task in manufacturing inspection. Reconstructive anomaly detection methods restore the normal appearance of an object, ideally modifying only the anomalous regions. However, previously commonly used reconstruction-based architecture always struggles with over-generalization and over-fitting problems, leading to poor reconstruction performance on real defective samples. In this study, we propose a more general denoising autoencoder, by introducing a feature hierarchy design to address these challenges in unsupervised anomaly detection. In particular, we operate feature transformation in the latent space to cope with the robustness of unseen anomalies in reality. Furthermore, the method enhances the discriminative capability of the model by focusing on multiple knowledge, including pixel color, HOG feature, and deep features. Additionally, a feature alignment module is proposed to manage the varied sizes and morphologies of features. Experiments conducted on both the MVTec AD dataset and the VisA dataset demonstrate that our FeatDAE significantly outperforms existing methods, achieving state-of-the-art results with high efficiency.