How to Reduce Anomaly Detection in Images to Anomaly Detection in Noise

Ehret, Thibaud; Davy, Axel; Delbracio, Mauricio; Morel, Jean‐Michel

doi:10.5201/ipol.2019.263

Cited by 9 publications

(6 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As previously mentioned, the a contrario approach is based on the same principle, but differs from the others in the sense that it allows to derive detection thresholds by estimating and controlling the expected number of occurrences of an event under the background model. An excellent example is the work by Ehret et al [15], in which they eliminate the self-similarity of the image by averaging the most frequent patches using Non Local Means [28] , and perform a detection using the NFA score over the residual image, which is formed by noise and anomalies. In this way, the general background modeling problem gets reduced to a noise modeling problem, making the algorithm to work with any kind of background.…”

Section: Related Workmentioning

confidence: 99%

“…Therefore, what we need to evaluate is the probability of that arrangement being generated by the background model. Another characteristic that makes the a contrario framework different and particularly useful, is that it automatically fixes detection thresholds that control the number of false alarms (NFA) [3], [15], [25], [26], allowing not only to detect rare events in very diverse backgrounds but also to associate a rareness score. This score has a clear statistical meaning: it is an estimate of the number of occurrences of an observed event if it was produced by the background model.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Contrario multi-scale anomaly detection method for industrial quality inspection

Tailanián¹,

Musé²,

Pardo³

2022

Preprint

View full text Add to dashboard Cite

Anomalies can be defined as any non-random structure which deviates from normality. Anomaly detection methods reported in the literature are numerous and diverse, as what is considered anomalous usually varies depending on particular scenarios and applications. In this work we propose an a contrario framework to detect anomalies in images applying statistical analysis to feature maps obtained via convolutions. We evaluate filters learned from the image under analysis via patch PCA, Gabor filters and the feature maps obtained from a pre-trained deep neural network (Resnet). The proposed method is multiscale and fully unsupervised and is able to detect anomalies in a wide variety of scenarios. While the end goal of this work is the detection of subtle defects in leather samples for the automotive industry, we show that the same algorithm achieves state-of-theart results in public anomalies datasets.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Contrario multi-scale anomaly detection method for industrial quality inspection

Tailanián¹,

Musé²,

Pardo³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The results have been compared with the ones achieved using a Residual Image Algorithm that compares a newly acquired image with a saved reference, similar to the method in [88]. For this algorithm, the Evaluator that has been used follows the same rules of the one used for the BBS-ESN, and it is summarized in Figure 10.…”

Section: Comparison With Residual Image Algorithmmentioning

confidence: 99%

Online Learning of Oil Leak Anomalies in Wind Turbines with Block-Based Binary Reservoir

et al. 2021

View full text Add to dashboard Cite

The focus of this work is to design a deeply quantized anomaly detector of oil leaks that may happen at the junction between the wind turbine high-speed shaft and the external bracket of the power generator. We propose a block-based binary shallow echo state network (BBS-ESN) architecture belonging to the reservoir computing (RC) category and, as we believe, it also extends the extreme learning machines (ELM) domain. Furthermore, BBS-ESN performs binary block-based online training using fixed and minimal computational complexity to achieve low power consumption and deployability on an off-the-shelf micro-controller (MCU). This has been achieved through binarization of the images and 1-bit quantization of the network weights and activations. 3D rendering has been used to generate a novel publicly available dataset of photo-realistic images similar to those potentially acquired by image sensors on the field while monitoring the junction, without and with oil leaks. Extensive experimentation has been conducted using a STM32H743ZI2 MCU running at 480 MHz and the results achieved show an accurate identification of anomalies, with a reduced computational cost per image and memory occupancy. Based on the obtained results, we conclude that BBS-ESN is feasible on off-the-shelf 32 bit MCUs. Moreover, the solution is also scalable in the number of image cameras to be deployed and to achieve accurate and fast oil leak detections from different viewpoints.

show abstract

“…As previously mentioned, the a contrario approach is based on the same principle, but differs from the others in the sense that it allows to derive detection thresholds by estimating and controlling the expected number of occurrences of an event under the background model. An excellent example is the work by Ehret et al [15], in which they eliminate the self-similarity of the image by averaging the most frequent patches using Non Local Means, and perform a detection using the NFA score over the residual image, which is formed by noise and anomalies. In this way, the general background modeling problem gets reduced to a noise modeling problem, making the algorithm to work with any kind of background.…”

Section: Related Workmentioning

confidence: 99%