The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection

Bergmann, Paul; Batzner, Kilian; Fauser, Michael; Sattlegger, David; Steger, Carsten

doi:10.1007/s11263-020-01400-4

Cited by 223 publications

(143 citation statements)

References 30 publications

Supporting

Mentioning

123

Contrasting

Unclassified

Order By: Relevance

“…Nine publicly available image datasets with real anomalies from diverse application domains are used. These include five benchmarks for identifying visual defects or microcracks on different object surfaces: MVTec AD 1 , AITEX 2 , SDD 3 [61] is a real-world industrial defect inspection dataset that contains 5,354 high-resolution images covering five types of texture defects and ten types of object defects. Each type of defect contains one to eight fine-grained types of defects.…”

Section: Datasetsmentioning

confidence: 99%

Deep Anomaly Detection with Deviation Networks

Pang

Shen

Hengel

2019

Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

263

168

View full text Add to dashboard Cite

Although deep learning has been applied to successfully address many data mining problems, relatively limited work has been done on deep learning for anomaly detection. Existing deep anomaly detection methods, which focus on learning new feature representations to enable downstream anomaly detection methods, perform indirect optimization of anomaly scores, leading to data-inefficient learning and suboptimal anomaly scoring. Also, they are typically designed as unsupervised learning due to the lack of large-scale labeled anomaly data. As a result, they are difficult to leverage prior knowledge (e.g., a few labeled anomalies) when such information is available as in many real-world anomaly detection applications.This paper introduces a novel anomaly detection framework and its instantiation to address these problems. Instead of representation learning, our method fulfills an end-to-end learning of anomaly scores by a neural deviation learning, in which we leverage a few (e.g., multiple to dozens) labeled anomalies and a prior probability to enforce statistically significant deviations of the anomaly scores of anomalies from that of normal data objects in the upper tail. Extensive results show that our method can be trained substantially more data-efficiently and achieves significantly better anomaly scoring than state-of-the-art competing methods.

show abstract

Section: Datasetsmentioning

confidence: 99%

Deep Anomaly Detection with Deviation Networks

Pang

Shen

Hengel

2019

Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

263

168

View full text Add to dashboard Cite

show abstract

“…The 'supervised one' and 'semi-supervised one' require a labeled data set which includes both normal and abnormal data. On the other hand, many studies [34], [35] rely on the 'unsupervised one' because it works with unlabeled data set. When most of the instances in the data set are normal, it can detect anomalies by just checking how much a data deviates from the normal region.…”

Section: E Confidence Level ( ) Calculationmentioning

confidence: 99%

“…2) were all adopted to this experiment. At the end of experiment, participants were asked to respond questionnaire of 'mental demand,' which was scaled to 0-20 by reference to NASA task load index [34]. As a result, the system detected drowsy more than two times for some participants, and the total detection number of times was nine.…”

Section: A Drowsiness Detectionmentioning

confidence: 99%

Toward Health–Related Accident Prevention: Symptom Detection and Intervention Based on Driver Monitoring and Verbal Interaction

Hayashi

Kamezaki

Sugano

2021

IEEE Open J. Intell. Transp. Syst.

View full text Add to dashboard Cite

Professional drivers are required to safely transport passengers and/or properties of customers to their destinations, so they must keep being mentally and physically healthy. Health problems will largely affect driving performance and sometimes cause loss of consciousness, which results in injury, death, and heavy compensation. Conventional systems can detect the loss of consciousness or urgently stop the vehicle to prevent accidents, but detection of symptoms of diseases and providing support before the driver loses consciousness is more reasonable. It is challenging to earlier detect symptoms with high confidence. Toward solving these problems, we propose a new method with a multi-sensor based driver monitoring system to detect cues of symptoms quickly and a verbal interaction system to confirm the internal state of the driver based on the monitoring results to reduce false positives. There is almost no data that records abnormal conditions while driving and tests with unhealthy participants are dangerous and ethically unacceptable, so we developed a system with pseudo-symptom data and did outlier detection only with normal driving data. From data collection experiments, we defined the confidence level derived from cue signs. The results of evaluation experiments showed that the proposed system worked well in pseudo headache and drowsiness detection scenarios. We found that signs of drowsiness varied with individual drivers, so the multi-sensor based driver monitoring system was proved to be effective. Moreover, we found that there were individual differences in how the cue signs appeared, so we can propose an online re-training method to make the system adapt to individual drivers.

show abstract

“…However, it required additional feature fusion modules. Bergman et al [21] proposed the largest industrial anomaly detection dataset, MVTec, and designed a method based on knowledge distillation for feature comparison. In order to improve the accuracy of anomaly localization, the input image in the model had to be divided into patches, which greatly increased the time and costs.…”

Section: Related Workmentioning

confidence: 99%

Automatic Unsupervised Fabric Defect Detection Based on Self-Feature Comparison

et al. 2021

View full text Add to dashboard Cite

Due to the huge demand for textile production in China, fabric defect detection is particularly attractive. At present, an increasing number of supervised deep-learning methods are being applied in surface defect detection. However, the annotation of datasets in industrial settings often depends on professional inspectors. Moreover, the methods based on supervised learning require a lot of annotation, which consumes a great deal of time and costs. In this paper, an approach based on self-feature comparison (SFC) was employed that accurately located and segmented fabric texture images to find anomalies with unsupervised learning. The SFC architecture contained the self-feature reconstruction module and the self-feature distillation. Accurate fiber anomaly location and segmentation were generated based on these two modules. Compared with the traditional methods that operate in image space, the comparison of feature space can better locate the anomalies of fiber texture surfaces. Evaluations were performed on the three publicly available databases. The results indicated that our method performed well compared with other methods, and had excellent defect detection ability in the collected textile images. In addition, the visual results showed that our results can be used as a pixel-level candidate label.

show abstract

The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection

Cited by 223 publications

References 30 publications

Deep Anomaly Detection with Deviation Networks

Deep Anomaly Detection with Deviation Networks

Toward Health–Related Accident Prevention: Symptom Detection and Intervention Based on Driver Monitoring and Verbal Interaction

Automatic Unsupervised Fabric Defect Detection Based on Self-Feature Comparison

Contact Info

Product

Resources

About