Focus Your Distribution: Coarse-to-Fine Non-Contrastive Learning for Anomaly Detection and Localization

Zheng, Ye; Wang, Xiang; Deng, Rui; Tianpeng, Bao,; Zhao, Rui; Wu, Liwei

doi:10.48550/arxiv.2110.04538

Cited by 4 publications

(6 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For MVTec AD dataset for 2D task, we report the performance comparison of Glance [31], DRAEM [35], DFR [32], R-D [9], PaDim [8], P-SVDD [33], FYD [37], SPADE [7], PANDA [20], CutPaste [18], NSA [28], CFlow [14], FastFlow [34], PatchCore [22] in terms of the image-level and pixel-level metrics. The inference efficiencies of some of these methods are also provided.…”

Section: Resultsmentioning

confidence: 99%

“…For the feature extractor, existing approaches usually adopt the CNN based or ViT based model [7,8,22,34], such as the ResNet [15] or ViT [11] trained on the Ima-geNet. Furthermore, to learn the domain-specific semantic vectors for images, a series of methods [18,20,37] employed self-supervised learning to achieve the ImageNet pretrained model adaption. These methods train the feature extractor by proposing different pretext tasks.…”

Section: Representation-based Methodsmentioning

confidence: 99%

“…However, the representation of created negative irregularities usually does not overlap with real-world anomalies [18], which limits the generalization potential of these methods in inference processes. FYD [37] takes into account spatial consistency on image representations to detect and localize the fine-grained defects. It proposes a non-contrastive learning method to train the model to avoid constructing negative samples and learn a dense and com-pact distribution from normal images with a coarse-to-fine alignment process.…”

Section: Representation-based Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Benchmarking Unsupervised Anomaly Detection and Localization

Zheng¹,

Wang²,

Qian³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Unsupervised anomaly detection and localization, as of one the most practical and challenging problems in computer vision, has received great attention in recent years. From the time the MVTec AD dataset was proposed to the present, new research methods that are constantly being proposed push its precision to saturation. It is the time to conduct a comprehensive comparison of existing methods to inspire further research. This paper extensively compares 13 papers in terms of the performance in unsupervised anomaly detection and localization tasks, and adds a comparison of inference efficiency previously ignored by the community. Meanwhile, analysis of the MVTec AD dataset are also given, especially the label ambiguity that affects the model fails to achieve full marks. Moreover, considering the proposal of the new MVTec 3D-AD dataset, this paper also conducts experiments using the existing state-of-the-art 2D methods on this new dataset, and reports the corresponding results with analysis.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Representation-based Methodsmentioning

confidence: 99%

Section: Representation-based Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Benchmarking Unsupervised Anomaly Detection and Localization

Zheng¹,

Wang²,

Qian³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…With a coarse-to-fine alignment technique, FYD method [101] seeks to learn dense and compact distribution of normal images. In both picture and feature levels, the coarse alignment stage normalizes the pixel-level position of objects.…”

Section: Representation Based Methodsmentioning

confidence: 99%

A Survey on Unsupervised Industrial Anomaly Detection Algorithms

Cui¹,

Liu²,

Lian³

2022

Preprint

View full text Add to dashboard Cite

In line with the development of Industry 4.0, more and more attention is attracted to the field of surface defect detection. Improving efficiency as well as saving labor costs has steadily become a matter of great concern in industry field, where deep learning-based algorithms performs better than traditional vision inspection methods in recent years. While existing deep learning-based algorithms are biased towards supervised learning, which not only necessitates a huge amount of labeled data and a significant amount of labor, but it is also inefficient and has certain limitations. In contrast, recent research shows that unsupervised learning has great potential in tackling above disadvantages for visual anomaly detection. In this survey, we summarize current challenges and provide a thorough overview of recently proposed unsupervised algorithms for visual anomaly detection covering five categories, whose innovation points and frameworks are described in detail. Meanwhile, information on publicly available datasets containing surface image samples are provided. By comparing different classes of methods, the advantages and disadvantages of anomaly detection algorithms are summarized. It is expected to assist both the research community and industry in developing a broader and cross-domain perspective.

show abstract

“…State-of-the-art It is worth mentioning that more recent methods have claimed better results on the same benchmarks used in this work. For instance, at least 5 papers [27,18,34,33,15] claim to have a mean ROC-AUC above 98% on the leaderboard for anomaly segmentation ("pixel-wise AD") on MVTec-AD in Papers with Code [8]. Unfortunately, we did not have the time to fully verify the experimental conditions in the sources but this serves as proxy evidence to take these results with consideration.…”

Section: Beyond the Papermentioning

confidence: 99%

[Reproducibility Report] Explainable Deep One-Class Classification

Bertoldo¹,

Decencière²

2022

Preprint

View full text Add to dashboard Cite

Scope of ReproducibilityLiznerski et al. [23] proposed Fully Convolutional Data Description (FCDD), an explainable version of the Hypersphere Classifier (HSC) to directly address image anomaly detection (AD) and pixel-wise AD without any post-hoc explainer methods. The authors claim that FCDD achieves results comparable with the state-of-the-art in sample-wise AD on Fashion-MNIST and CIFAR-10 and exceeds the state-of-the-art on the pixel-wise task on MVTec-AD. They also give evidence to show a clear improvement by using few (1 up to 8) real anomalous images in MVTec-AD for supervision at the pixel level. Finally, a qualitative study with horse images on PASCAL-VOC shows that FCDD can intrinsically reveal spurious model decisions by providing built-in anomaly score heatmaps. MethodologyWe have reproduced the quantitative results in the main text of [23] except for the performance on ImageNet: samplewise AD on Fashion-MNIST and CIFAR-10, and pixel-wise AD on MVTec-AD. We used the author's code with GPUs NVIDIA TITAN X and NVIDIA TITAN Xp. A more detailed look into FCDD's performance variability is presented, and a Critical Difference (CD) diagram is proposed as a more appropriate tool to compare methods over the datasets in MVTec-AD. Finally, we study the generalization power of the unsupervised FCDD during training. ResultsAll per-class performances (in terms of Area Under the ROC Curve (ROC-AUC) [31]) announced in the paper were replicated with absolute difference of at most 2% and below 1% on average, confirming the paper's claims. We report the experiments' GPU and CPU memory requirements and their average training time. Our analyses beyond the paper's scope show that claiming to "exceed the state-of-the-art" should be considered with care, and evidence is given to argue that the pixel-wise unsupervised FCDD could narrow the gap with its semi-supervised version.

show abstract

Focus Your Distribution: Coarse-to-Fine Non-Contrastive Learning for Anomaly Detection and Localization

Cited by 4 publications

References 29 publications

Benchmarking Unsupervised Anomaly Detection and Localization

Benchmarking Unsupervised Anomaly Detection and Localization

A Survey on Unsupervised Industrial Anomaly Detection Algorithms

[Reproducibility Report] Explainable Deep One-Class Classification

Contact Info

Product

Resources

About