FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flows

By Jiawei Yu1, Ye Zheng, Xiang Wang, Wei Li, Yushuang Wu, Rui Zhao, Liwei Wu. arXiv.org, 2021

475 citations recorded by the source corpus.

Unsupervised anomaly detection and localization is crucial to the practical application when collecting and labeling sufficient anomaly data is infeasible. Most existing representation-based approaches extract normal image features with a deep convolutional neural network and characterize the corresponding distribution through non-parametric distribution estimation methods. The anomaly score is calculated by measuring the distance between the feature of the test image and the estimated distribution. However, current methods can not effectively map image features to a tractable base distribution and ignore the relationship between local and global features which are important to identify anomalies. To this end, we propose FastFlow implemented with 2D normalizing flows and use it as the probability distribution estimator. Our FastFlow can be used as a plug-in module with arbitrary deep feature extractors such as ResNet and vision transformer for unsupervised anomaly detection and localization. In training phase, FastFlow learns to transform the input visual feature into a tractable distribution and obtains the likelihood to recognize anomalies in inference phase. Extensive experimental results on the MVTec AD dataset show that FastFlow surpasses previous state-of-the-art methods in terms of accuracy and inference efficiency with various backbone networks. Our approach achieves 99.4% AUC in anomaly detection with high inference efficiency.

Subfields: Artificial Intelligence, Computer Networks and Communications, Epidemiology

View Paper: https://www.semanticscholar.org/paper/11709bfadfd6bbb371f4077bccb7c26d93c39cdd

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Onkar Dabeer
2. Kilian Batzner
3. Jiulong Shan
4. Dongqing Zhang
5. Yang Zou
6. Xi Li
7. Lars Heckler
8. Taewan Kim

In Computer Networks and Communications

1. Onkar Dabeer
2. Kilian Batzner
3. Dongqing Zhang
4. Yang Zou
5. Xi Li
6. Lars Heckler
7. Avinash Ravichandran
8. Rebecca König

In Epidemiology

1. Kilian Batzner
2. Lars Heckler
3. Rebecca König

Citations in the corpus

• EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies (2023, methodology): “• We propose an efficient network architecture to speed up feature extraction by an order of magnitude in comparison to the feature extractors used by recent meth-ods [47,50,64].”
• Deep Industrial Image Anomaly Detection: A Survey (2023, unknown): “FastFlow [57] introduces an alternate stacking of large and small convolution kernels in the NF module to model global and local distribution efficiently.”
• WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation (2023, background): “While the MVTec-AD benchmark [3] is saturated by several works [31, 47, 50], their specific application is hindered due to their unscalable full-normal-shot setup.”
• DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection (2022, background): “Since the assumption of Gaussian distribution is too strict, some recent works borrow ideas from normalizing flow, by projecting an arbitrary distribution to a Gaussian distribution to approximate the density of any distributions [13, 36].”