FAIR: Frequency-aware Image Restoration for Industrial Visual Anomaly Detection
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916651972165632 |
|---|---|
| author | Liu, Tongkun Li, Bing Du, Xiao Jiang, Bingke Geng, Leqi Wang, Feiyang Zhao, Zhuo |
| author_facet | Liu, Tongkun Li, Bing Du, Xiao Jiang, Bingke Geng, Leqi Wang, Feiyang Zhao, Zhuo |
| contents | Image reconstruction-based anomaly detection models are widely explored in industrial visual inspection. However, existing models usually suffer from the trade-off between normal reconstruction fidelity and abnormal reconstruction distinguishability, which damages the performance. In this paper, we find that the above trade-off can be better mitigated by leveraging the distinct frequency biases between normal and abnormal reconstruction errors. To this end, we propose Frequency-aware Image Restoration (FAIR), a novel self-supervised image restoration task that restores images from their high-frequency components. It enables precise reconstruction of normal patterns while mitigating unfavorable generalization to anomalies. Using only a simple vanilla UNet, FAIR achieves state-of-the-art performance with higher efficiency on various defect detection datasets. Code: https://github.com/liutongkun/FAIR. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2309_07068 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | FAIR: Frequency-aware Image Restoration for Industrial Visual Anomaly Detection Liu, Tongkun Li, Bing Du, Xiao Jiang, Bingke Geng, Leqi Wang, Feiyang Zhao, Zhuo Computer Vision and Pattern Recognition Image reconstruction-based anomaly detection models are widely explored in industrial visual inspection. However, existing models usually suffer from the trade-off between normal reconstruction fidelity and abnormal reconstruction distinguishability, which damages the performance. In this paper, we find that the above trade-off can be better mitigated by leveraging the distinct frequency biases between normal and abnormal reconstruction errors. To this end, we propose Frequency-aware Image Restoration (FAIR), a novel self-supervised image restoration task that restores images from their high-frequency components. It enables precise reconstruction of normal patterns while mitigating unfavorable generalization to anomalies. Using only a simple vanilla UNet, FAIR achieves state-of-the-art performance with higher efficiency on various defect detection datasets. Code: https://github.com/liutongkun/FAIR. |
| title | FAIR: Frequency-aware Image Restoration for Industrial Visual Anomaly Detection |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2309.07068 |