A taxonomy of Unsupervised Industrial Anomaly Detection (UIAD) methods and datasets (updating).
Welcome to follow our papers "A survey on RGB, 3D, and multimodal approaches for unsupervised industrial image anomaly detection".
If you find any errors in our survey and resource repository, or if you have any suggestions, please feel free to contact us via email at: [email protected].
- [2025.03.19] 🔥🔥 Accepted by Information Fusion!
Dataset | Resource | Year | Type | Train | Test (good) | Test (anomaly) | Val | Total | Class | Anomaly Type | Modal Type |
---|---|---|---|---|---|---|---|---|---|---|---|
MVTec AD |
Data | 2019 | Real | 3629 | 467 | 1258 | - | 5354 | 15 | 73 | RGB |
BTAD |
Data | 2021 | Real | 1799 | 451 | 290 | - | 2540 | 3 | - | RGB |
MPDD |
Data | 2021 | Real | 888 | 176 | 282 | - | 1346 | 6 | - | RGB |
MVTec LOCO-AD |
Data | 2022 | Real | 1772 | 575 | 993 | 304 | 3644 | 5 | 89 | RGB |
VisA |
Data | 2022 | Real | 9621 | 0 | 1200 | - | 10821 | 12 | - | RGB |
GoodsAD |
Data | 2023 | Real | 3136 | 1328 | 1660 | - | 6124 | 6 | - | RGB |
MSC-AD |
- | 2023 | Real | 6480 | 2160 | 1080 | - | 9720 | 12 | 5 | RGB |
CID |
Data | 2024 | Real | 3900 | 33 | 360 | - | 4293 | 1 | 6 | RGB |
Real-IAD |
Data | 2024 | Real | 72840 | 0 | 78210 | - | 151050 | 30 | 8 | RGB |
RAD |
Data | 2024 | Real | 213 | 73 | 1224 | - | 1510 | 4 | - | RGB |
MIAD |
Data | 2023 | Synthetic | 70000 | 17500 | 17500 | - | 105000 | 7 | 13 | RGB |
MAD-Sim |
Data | 2023 | Synthetic | 4200 | 638 | 4951 | - | 9789 | 20 | 3 | RGB |
DTD-Synthetic |
Data | 2024 | Synthetic | 1200 | 357 | 947 | - | 2504 | 12 | - | RGB |
Dataset | Resource | Year | Type | Train | Test (good) | Test (anomaly) | Val | Total | Class | Anomaly Type | Modal Type |
---|---|---|---|---|---|---|---|---|---|---|---|
Real3D-AD |
data | 2023 | Real | 48 | 604 | 602 | - | 1254 | 12 | 3 | Point cloud |
Anomaly-ShapeNet |
data | 2023 | Synthetic | 208 | 780 | 943 | - | 1931 | 50 | 7 | Point cloud |
Name | Title | Publication | Year | Code | Paradigm |
---|---|---|---|---|---|
3D-ST | Anomaly Detection in 3D Point Clouds Using Deep Geometric Descriptors |
WACV | 2023 | - | Teacher-student architecture |
Reg3D-AD | Real3D-AD: A Dataset of Point Cloud Anomaly Detection |
NeurIPS | 2024 | Code | Memory bank |
Group3AD | Towards High-resolution 3D Anomaly Detection via Group-Level Feature Contrastive Learning |
ACM MM | 2024 | Code | Memory bank |
PointCore | PointCore: Efficient Unsupervised Point Cloud Anomaly Detector Using Local-Global Features |
- | 2024 | - | Memory bank |
R3D-AD | R3D-AD: Reconstruction via Diffusion for 3D Anomaly Detection |
ECCV | 2024 | - | Reconstruction |
Dataset | Resource | Year | Type | Train | Test (good) | Test (anomaly) | Val | Total | Class | Anomaly Type | Modal Type |
---|---|---|---|---|---|---|---|---|---|---|---|
MVTec 3D-AD |
data | 2021 | Real | 2656 | 294 | 948 | 294 | 4147 | 10 | 41 | RGB & Point cloud |
PD-REAL |
data | 2023 | Real | 2399 | 300 | 530 | 300 | 3529 | 15 | 6 | RGB & Point cloud |
MulSen-AD |
data | 2024 | Real | 1391 | 150 | 494 | - | 2035 | 15 | 14 | RGB & Infrared & Point cloud |
Eyecandies |
data | 2022 | Synthetic | 10000 | 2250 | 2250 | 1000 | 15500 | 10 | - | RGB & Depth |
If you find this paper and repository useful, please cite our paper:
@article{lin2025survey,
title={A survey on RGB, 3D, and multimodal approaches for unsupervised industrial image anomaly detection},
author={Lin, Yuxuan and Chang, Yang and Tong, Xuan and Yu, Jiawen and Liotta, Antonio and Huang, Guofan and Song, Wei and Zeng, Deyu and Wu, Zongze and Wang, Yan and others},
journal={Information Fusion},
pages={103139},
year={2025},
publisher={Elsevier}
}