Self-Supervised Training with Autoencoders for Visual Anomaly Detection

Deep convolutional autoencoders provide an effective tool for learning non-linear dimensionality reduction in an unsupervised way. Recently, they have been used for the task of anomaly detection in the visual domain. By optimising for the reconstruction error using anomaly-free examples, the common belief is that a corresponding network should fail to accurately reconstruct anomalous regions in the application phase. This goal is typically addressed by controlling the capacity of the network by either reducing the size of the bottleneck layer or enforcing sparsity constraints on its activations. However, neither of these techniques does explicitly penalize reconstruction of anomalous signals often resulting in poor detection. We tackle this problem by adapting a self-supervised learning regime, which allows to use discriminative information during training focusing on the data manifold by means of a modified reconstruction error. This regularizes the model to produce locally consistent reconstructions, while replacing irregularities by acting as a filter for anomalous patterns. In contrast to related approaches, inference with our method is very efficient during training and prediction processing the entire input image in one single step. Our experiments on the MVTec AD dataset demonstrate high recognition and localization performance of the proposed method. On the texture-subset, in particular, our approach consistently outperforms a bunch of recent anomaly detection methods by a big margin.