Visual Anomaly Detection on ViCoS Lab

Analiza robustnosti globokih nenadzorovanih metod za detekcijo vizualnih anomalij

Mon, 01 Jan 0001 00:00:00 +0000

Unsupervised generative methods have recently attracted significant attention in the field of industrial visual anomaly detection, mainly owing to their ability to learn from non anomalous data withouth requiring anomalous samples and pixel-level labels, which are costly to obtain. An assumption that anomalous data are always correctly identified and consequently removed from the training set underlies all of the generative methods. In practice, however, correctly identifying every single anomalous image can either be very costly to do or it can not be done at all due to the nature of the problem. In this paper, we analyze how robust some of the recently proposed generative methods for anomaly detection are, by introducing anomalous data in the training process. Our analysis covers 3 methods and 4 datasets with 8 categories in total, and we conclude that while some of the methods are more robust than others, introducing a minor percentage of anomalous data in the training set does not significantly deteriorate the performance.

DRAEM -- A discriminatively trained reconstruction embedding for surface anomaly detection

Mon, 01 Jan 0001 00:00:00 +0000

Visual surface anomaly detection aims to detect local image regions that significantly deviate from normal appearance. Recent surface anomaly detection methods rely on generative models to accurately reconstruct the normal areas and to fail on anomalies. These methods are trained only on anomaly-free images, and often require hand-crafted post-processing steps to localize the anomalies, which prohibits optimizing the feature extraction for maximal detection capability. In addition to reconstructive approach, we cast surface anomaly detection primarily as a discriminative problem and propose a discriminatively trained reconstruction anomaly embedding model (DRAEM). The proposed method learns a joint representation of an anomalous image and its anomaly-free reconstruction, while simultaneously learning a decision boundary between normal and anomalous examples. The method enables direct anomaly localization without the need for additional complicated post-processing of the network output and can be trained using simple and general anomaly simulations. On the challenging MVTec anomaly detection dataset, DRAEM outperforms the current state-of-the-art unsupervised methods by a large margin and even delivers detection performance close to the fully-supervised methods on the widely used DAGM surface-defect detection dataset, while substantially outperforming them in localization accuracy.

Reconstruction by inpainting for visual anomaly detection

Mon, 01 Jan 0001 00:00:00 +0000

Visual anomaly detection addresses the problem of classification or localization of regions in an image that deviate from their normal appearance. A popular approach trains an auto-encoder on anomaly-free images and performs anomaly detection by calculating the difference between the input and the reconstructed image. This approach assumes that the auto-encoder will be unable to accurately reconstruct anomalous regions. But in practice neural networks generalize well even to anomalies and reconstruct them sufficiently well, thus reducing the detection capabilities. Accurate reconstruction is far less likely if the anomaly pixels were not visible to the auto-encoder. We thus cast anomaly detection as a self-supervised reconstruction-by-inpainting problem. Our approach (RIAD) randomly removes partial image regions and reconstructs the image from partial inpaintings, thus addressing the drawbacks of auto-enocoding methods. RIAD is extensively evaluated on several benchmarks and sets a new state-of-the art on a recent highly challenging anomaly detection benchmark.