Anomaly Detection of GAN Industrial Image Based on Attention Feature Fusion - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Dec 29;23(1):355.
doi: 10.3390/s23010355.

Anomaly Detection of GAN Industrial Image Based on Attention Feature Fusion

Lin Zhang  1 , Yang Dai  1 , Fuyou Fan  2 , Chunlin He  1
Affiliations

Anomaly Detection of GAN Industrial Image Based on Attention Feature Fusion

Lin Zhang et al. Sensors (Basel). .

Abstract

As life becomes richer day by day, the requirement for quality industrial products is becoming greater and greater. Therefore, image anomaly detection on industrial products is of significant importance and has become a research hotspot. Industrial manufacturers are also gradually intellectualizing how product parts may have flaws and defects, and that industrial product image anomalies have characteristics such as category diversity, sample scarcity, and the uncertainty of change; thus, a higher requirement for image anomaly detection has arisen. For this reason, we proposed a method of industrial image anomaly detection that applies a generative adversarial network based on attention feature fusion. For the purpose of capturing richer image channel features, we added attention feature fusion based on an encoder and decoder, and through skip-connection, this performs the feature fusion for the encode and decode vectors in the same dimension. During training, we used random cut-paste image augmentation, which improved the diversity of the datasets. We displayed the results of a wide experiment, which was based on the public industrial detection MVTec dataset. The experiment illustrated that the method we proposed has a higher level AUC and the overall result was increased by 4.1%. Finally, we realized the pixel level anomaly localization of the industrial dataset, which illustrates the feasibility and effectiveness of this method.

Keywords: anomaly detection; attention feature fusion; generative adversarial network; image augmentation.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
On the left, (A) is the image enhancement process and (B) is the image reconstruction process. Generators learn only normal image features. On the right, (A) indicates that the image has not been data enhanced. (B) is the reconstruction process of the test image. (C) represents the absolute residual between the reconstructed image and the test image, and the residual image is obtained. Finally, the thermal map is generated according to the residual image, and a 0.7 times thermal map is superimposed on the 0.3 times abnormal image to complete the abnormal location of the image.
Figure 2
Figure 2
The architecture of the network used in our method.
Figure 3
Figure 3
Attention feature fusion: the blue area represents the multiple scale channel attention module, ⊕ denotes the broadcasting addition, and ⊗ denotes the element-wise multiplication.
Figure 4
Figure 4
Examples of the image augmentation effect. This method randomly cuts a small rectangular area (the red rectangular boxes) and pastes these into random positions.
Figure 5
Figure 5
Normal samples and samples with surface defects in the MVTec dataset. The area in the red box contains the surface defect of each product.
Figure 6
Figure 6
Visualization of the AUC values of the method we proposed and the other six methods.
Figure 7
Figure 7
The algorithm frame diagram of the encode–decode with a skip-connection.
Figure 8
Figure 8
The detection results of all of the categories. From left to right: anomaly images, reconstructed images, residual image, heat map, and ground truth.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Pang G., Shen C., Cao L., Hengel A.V.D. Deep Learning for Anomaly Detection: A Review. ACM Comput. Surv. 2021;54:1–38. doi: 10.1145/3439950. - DOI
    1. Ahmed M., Mahmood A.N., Islam R. A survey of anomaly detection techniques in financial domain. Future Gener. Comput. Syst. 2016;55:278–288. doi: 10.1016/j.future.2015.01.001. - DOI
    1. Atefeh F., Khreich W. A Survey of Techniques for Event Detection in Twitter. Comput. Intell. 2015;31:132–164. doi: 10.1111/coin.12017. - DOI
    1. Abdallah A., Maarof M.A., Zainal A. Fraud detection system: A survey. J. Netw. Comput. Appl. 2016;68:90–113. doi: 10.1016/j.jnca.2016.04.007. - DOI
    1. Alvarenga T.A., Carvalho A.L., Honorio L.M., Cerqueira A.S., Filho L.M.A., Nobrega R.A. Detection and classification system for rail surface defects based on Eddy current. Sensors. 2021;21:7937. doi: 10.3390/s21237937. - DOI - PMC - PubMed