Home

Awesome

Awesome Industrial Anomaly Detection Awesome

We discuss public datasets and related studies in detail. Welcome to read our paper and make comments.

Deep Industrial Image Anomaly Detection: A Survey (Machine Intelligence Research)

IM-IAD: Industrial Image Anomaly Detection Benchmark in Manufacturing [TCYB 2024][code][中文]

We will keep focusing on this field and updating relevant information.

Keywords: anomaly detection, anomaly segmentation, industrial image, defect detection

[Main Page] [Survey] [Benchmark] [Result]

🔥🔥🔥 Contributions to our repository are welcome. Feel free to categorize the papers.


🔥🔥🔥 Which MLLM performs best in industrial anomaly detection? Please refer to our recent research, which evaluates state-of-the-art models, including GPT-4o, Gemini-1.5, LLaVA-Next, and InternVL.

[2024.10.16] We are proud to announce the launch of MMAD, the first-ever comprehensive benchmark for Multimodal Large Language Models in Industrial Anomaly Detection! 🌟 [Paper] [Code] [Data]


Table of Contents

SOTA methods with code

TitleVenueDateCodetopic
Star <br> Anomaly Detection via Reverse Distillation from One-Class Embedding <br>CVPR2022GithubTeacher-Student
Star <br> Revisiting Reverse Distillation for Anomaly Detection <br>CVPR2023GithubTeacher-Student
Star <br> SimpleNet: A Simple Network for Image Anomaly Detection and Localization <br>CVPR2023GithubOne-Class-Classification
Star <br> Real-time unsupervised anomaly detection with localization via conditional normalizing flows <br>WACV2022GithubDistribution Map
Star <br> PyramidFlow: High-Resolution Defect Contrastive Localization using Pyramid Normalizing Flow <br>CVPR2023GithubDistribution Map
Star <br> Towards total recall in industrial anomaly detection <br>CVPR2022GithubMemory-bank
Star <br> PNI: Industrial Anomaly Detection using Position and Neighborhood Information <br>ICCV2023GithubMemory-bank
Star <br> Draem-a discriminatively trained reconstruction embedding for surface anomaly detection <br>ICCV2021GithubReconstruction-based
Star <br> DSR: A dual subspace re-projection network for surface anomaly detection <br>ECCV2022GithubReconstruction-based
Star <br> Omni-frequency Channel-selection Representations for Unsupervised Anomaly Detection <br>TIP2023GithubReconstruction-based
Star <br> RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detection <br>CVPR2024GithubReconstruction-based
Star <br> Registration based few-shot anomaly detection <br>ECCV2022GithubFew Shot
Star <br> AnomalyGPT: Detecting Industrial Anomalies using Large Vision-Language Models <br>AAAI2024GithubFew Shot
Star <br> Catching Both Gray and Black Swans: Open-set Supervised Anomaly Detection <br>CVPR2022GithubFew abnormal samples
Star <br> Explicit Boundary Guided Semi-Push-Pull Contrastive Learning for Supervised Anomaly Detection <br>CVPR2023GithubFew abnormal samples
Star <br> Deep one-class classification via interpolated gaussian descriptor <br>AAAI2022GithubNoisy AD
Star <br> SoftPatch: Unsupervised Anomaly Detection with Noisy Data <br>NeurIPS2022GithubNoisy AD
Star <br> Inter-Realization Channels: Unsupervised Anomaly Detection Beyond One-Class Classification <br>ICCV2023GithubNoisy AD
Star <br> Unsupervised Continual Anomaly Detection with Contrastively-learned Prompt <br>AAAI2024GithubContinual AD
Star <br> A Unified Model for Multi-class Anomaly Detection <br>NeurIPS2022GithubMulti-class unified
Star <br> Hierarchical Vector Quantized Transformer for Multi-class Unsupervised Anomaly Detection <br>NeurIPS2023GithubMulti-class unified
Star <br> Multimodal Industrial Anomaly Detection via Hybrid Fusion <br>CVPR2023GithubRGBD
Star <br> Real3D-AD: A Dataset of Point Cloud Anomaly Detection <br>NeurIPS2023GithubPoint Cloud
Star <br> AnoVL: Adapting Vision-Language Models for Unified Zero-shot Anomaly Localization <br>arxiv2023GithubZero Shot
Star <br> Segment Any Anomaly without Training via Hybrid Prompt Regularization <br>arxiv2023GithubZero Shot
Star <br> PSAD: Few Shot Part Segmentation Reveals Compositional Logic for Industrial Anomaly Detection <br>AAAI2024GithubLogical/Few Shot
Star <br> UniFormaly: Towards Task-Agnostic Unified Framework for Visual Anomaly Detection <br>arxiv2023GithubMulti-class unified

Recommended Benchmarks

TitleVenueDateCodetopic
Star <br> Anomalib: A Deep Learning Library for Anomaly Detection <br>ICIP2022GithubBenchmark
Star <br> IM-IAD: Industrial Image Anomaly Detection Benchmark in Manufacturing <br>TCYB2024GithubBenchmark
Star <br> ADer: A Comprehensive Benchmark for Multi-class Visual Anomaly Detection <br>arxiv2024GithubBenchmark
Star <br> MMAD: The First-Ever Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detection <br>arxiv2024GithubBenchmark

Recent research

NeurIPS 2024

ECCV 2024

ACM MM 2024

ICASSP 2024

CVPR 2024

ICLR 2024

AAAI 2024

WACV 2024

NeurIPS 2023

<!-- ## ICML 2023 + Shape-Guided Dual-Memory Learning for 3D Anomaly Detection [[ICML 2023]](https://openreview.net/forum?id=IkSGn9fcPz) + Fascinating Supervisory Signals and Where to Find Them: Deep Anomaly Detection with Scale Learning [[ICML 2023]](https://openreview.net/forum?id=V6PNBRWRil) ## ACM MM 2023 + EasyNet: An Easy Network for 3D Industrial Anomaly Detection [[ACM MM 2023]](https://arxiv.org/abs/2307.13925) ## ICCV 2023 + Remembering Normality: Memory-guided Knowledge Distillation for Unsupervised Anomaly Detection [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Gu_Remembering_Normality_Memory-guided_Knowledge_Distillation_for_Unsupervised_Anomaly_Detection_ICCV_2023_paper.pdf) + Unsupervised Surface Anomaly Detection with Diffusion Probabilistic Model [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Zhang_Unsupervised_Surface_Anomaly_Detection_with_Diffusion_Probabilistic_Model_ICCV_2023_paper.pdf) + PNI: Industrial Anomaly Detection using Position and Neighborhood Information [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Bae_PNI__Industrial_Anomaly_Detection_using_Position_and_Neighborhood_Information_ICCV_2023_paper.pdf)[[code]](https://github.com/wogur110/PNI_Anomaly_Detection) + Anomaly Detection using Score-based Perturbation Resilience [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Shin_Anomaly_Detection_using_Score-based_Perturbation_Resilience_ICCV_2023_paper.pdf) + Template-guided Hierarchical Feature Restoration for Anomaly Detection [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Guo_Template-guided_Hierarchical_Feature_Restoration_for_Anomaly_Detection_ICCV_2023_paper.pdf) + Focus the Discrepancy: Intra- and Inter-Correlation Learning for Image Anomaly Detection [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Yao_Focus_the_Discrepancy_Intra-_and_Inter-Correlation_Learning_for_Image_Anomaly_ICCV_2023_paper.pdf)[[code]](https://github.com/xcyao00/FOD) + Anomaly Detection under Distribution Shift [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Cao_Anomaly_Detection_Under_Distribution_Shift_ICCV_2023_paper.pdf)[[code]](https://github.com/mala-lab/ADShift) + FastRecon: Few-shot Industrial Anomaly Detection via Fast Feature Reconstruction [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Fang_FastRecon_Few-shot_Industrial_Anomaly_Detection_via_Fast_Feature_Reconstruction_ICCV_2023_paper.pdf)[[code]](https://github.com/FzJun26th/FastRecon) + Inter-Realization Channels: Unsupervised Anomaly Detection Beyond One-Class Classification [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/McIntosh_Inter-Realization_Channels_Unsupervised_Anomaly_Detection_Beyond_One-Class_Classification_ICCV_2023_paper.pdf)[[code]](https://github.com/DeclanMcIntosh/InReaCh) + Removing Anomalies as Noises for Industrial Defect Localization [[ICCV 2023]](https://openaccess.thecvf.com/content/ICCV2023/papers/Lu_Removing_Anomalies_as_Noises_for_Industrial_Defect_Localization_ICCV_2023_paper.pdf) -->

LLM related

<!-- ## CVPR 2023 + CVPR 2023 Tutorial on "Recent Advances in Anomaly Detection" [[CVPR Workshop 2023(mainly on video anomaly detection)]](https://sites.google.com/view/cvpr2023-tutorial-on-ad/)[[video]](https://www.youtube.com/watch?v=dXxrzWeybBo&feature=youtu.be) + Workshop on Vision-Based Industrial Inspection [[CVPR Workshop paper list 2023]](https://openaccess.thecvf.com/CVPR2023_workshops/VISION) + Visual Anomaly and Novelty Detection [[CVPR Workshop paper list 2023]](https://openaccess.thecvf.com/CVPR2023_workshops/VAND) + Revisiting Reverse Distillation for Anomaly Detection [[CVPR 2023]](https://openaccess.thecvf.com/content/CVPR2023/papers/Tien_Revisiting_Reverse_Distillation_for_Anomaly_Detection_CVPR_2023_paper.pdf) [[code]](https://github.com/tientrandinh/Revisiting-Reverse-Distillation) + OmniAL A unifiled CNN framework for unsupervised anomaly localization [[CVPR 2023]](https://openaccess.thecvf.com/content/CVPR2023/papers/Zhao_OmniAL_A_Unified_CNN_Framework_for_Unsupervised_Anomaly_Localization_CVPR_2023_paper.pdf) + Explicit Boundary Guided Semi-Push-Pull Contrastive Learning for Supervised Anomaly Detection [[CVPR 2023]](https://arxiv.org/abs/2207.01463)[[code]](https://github.com/xcyao00/BGAD) + DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection [[CVPR 2023]](https://arxiv.org/abs/2211.11317)[[code]](https://github.com/apple/ml-destseg) + Diversity-Measurable Anomaly Detection [[CVPR 2023]](https://arxiv.org/abs/2303.05047) + WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation [[CVPR 2023]](https://arxiv.org/abs/2303.14814) + SimpleNet: A Simple Network for Image Anomaly Detection and Localization [[CVPR 2023]](https://arxiv.org/abs/2303.15140)[[code]](https://github.com/DonaldRR/SimpleNet) + PyramidFlow: High-Resolution Defect Contrastive Localization using Pyramid Normalizing Flow [[CVPR 2023]](https://arxiv.org/abs/2303.02595)[[code]](https://github.com/gasharper/PyramidFlow) + Multimodal Industrial Anomaly Detection via Hybrid Fusion [[CVPR 2023]](https://arxiv.org/abs/2303.00601)[[code]](https://github.com/nomewang/M3DM) + Prototypical Residual Networks for Anomaly Detection and Localization [[CVPR 2023]](https://arxiv.org/abs/2212.02031)[[code]](https://github.com/xcyao00/PRNet) + SQUID: Deep Feature In-Painting for Unsupervised Anomaly Detection [[CVPR 2023]](https://arxiv.org/abs/2111.13495) + APRIL-GAN: A Zero-/Few-Shot Anomaly Classification and Segmentation Method for CVPR 2023 VAND Workshop Challenge Tracks 1&2: 1st Place on Zero-shot AD and 4th Place on Few-shot AD [[CVPR 2023 VAND Workshop Challenge]](https://arxiv.org/abs/2305.17382) -->

SAM segment anything

<!-- ## ICLR 2023 + Pushing the Limits of Fewshot Anomaly Detection in Industry Vision: Graphcore [[ICLR 2023]](https://openreview.net/pdf?id=xzmqxHdZAwO) + RGI: robust GAN-inversion for mask-free image inpainting and unsupervised pixel-wise anomaly detection [[ICLR 2023]](https://openreview.net/pdf?id=1UbNwQC89a) -->

Others

Medical (related)

Paper Tree (Classification of representative methods)

PaperTree

Timeline

Timeline

Paper list for industrial image anomaly detection

Related Survey, Benchmark, and Framework

2 Unsupervised AD

2.1 Feature-Embedding-based Methods

2.1.1 Teacher-Student

2.1.2 One-Class Classification (OCC)

2.1.3 Distribution-Map

2.1.4 Memory Bank

2.1.5 Vison Language AD

2.2 Reconstruction-Based Methods

2.2.1 Autoencoder (AE)

2.2.2 Generative Adversarial Networks (GANs)

2.2.3 Transformer

2.2.4 Diffusion Model

2.2.5 Others

2.3 Supervised AD

More Normal samples With (Less Abnormal Samples or Weak Labels)

More Abnormal Samples

3 Other Research Direction

3.1 Zero/Few-Shot AD

Zero-Shot AD

Few-Shot AD

3.2 Noisy AD

3.3 Anomaly Synthetic

3.4 RGBD AD

3.5 3D AD

3.6 Continual AD

3.7 Uniform/Multi-Class AD

3.8 Logical AD

Other settings

TTT binary segmentation

MoE with TTA

Adversary Attack

Defect Classification

4 Dataset

DatasetClassNormalAbnormalTotalAnnotation levelSourceTime
AITEX1140105245Segmentation maskRGB real2019
Anomaly-ShapeNet40--1600Point-level maskPoint-cloud syntheticCVPR,2024
BTAD3--2830Segmentation maskRGB real2021
CID140602334293Segmentation maskRGB real2024,TIM
DAGM10--11500Segmentation maskRGB synthetic2007
DEEPPCB1--1500Bounding boxRGB synthetic2019
DTD-Synthetic12---Segmentation maskRGB syntheticWACV,2024
Eyecandies1013250225015500Segmentation maskRGBD synthetic imageACCV,2022
Fabirc dataset1252550Segmentation maskRGB syntheticPR,2016
GDXray101940719407Bounding boxRGB real2016
IPAD16--597979ImageVideo real&synthetic2024
KolekrotSDD134752399Segmentation maskRGB realJIM,2019
KolekrotSDD2129793563335Segmentation maskRGB realCiI,2021
MIAD78750017500105000Segmentation maskRGB synthetic2023
MPDD610642821346Segmentation maskRGB realICUMT,2021
MTD19523921344Segmentation maskRGB realCASE,2018
MVTec AD15409612585354Segmentation maskRGB realCVPR,2019
MVTec 3D-AD1029049483852Segmentation maskRGB realVISAPP,2021
MVTec LOCO-AD523479933340Segmentation maskRGBD realIJCV,2022
NanoTwice154045Segmentation maskRGB realTII,2016
NEU surface defect1018001800Bounding boxRGB real2013
PAD205231490210133Segmentation maskRBG syntheticNeurIPS,2023
Real-IAD309972151329151050Segmentation maskRGB realCVPR,2024
Real3D-AD126526021254Point-level maskPoint-cloud realNeurIPS,2023
RSDD2--195Segmentation maskRGB real2017
Steel defect detection1--18076ImageRGB real2019
Steel tube dataset1034083408Bounding boxRGB real2021
VisA129621120010821Segmentation maskRGB realECCV,2022
RAD421312241224Segmentation maskRGB realCASE,2024

BibTex Citation

If you find this paper and repository useful, please cite our paper☺️.

@article{liu2024deep,
  title={Deep industrial image anomaly detection: A survey},
  author={Liu, Jiaqi and Xie, Guoyang and Wang, Jinbao and Li, Shangnian and Wang, Chengjie and Zheng, Feng and Jin, Yaochu},
  journal={Machine Intelligence Research},
  volume={21},
  number={1},
  pages={104--135},
  year={2024},
  publisher={Springer}
}

@article{xie2024iad,
  title={Im-iad: Industrial image anomaly detection benchmark in manufacturing},
  author={Xie, Guoyang and Wang, Jinbao and Liu, Jiaqi and Lyu, Jiayi and Liu, Yong and Wang, Chengjie and Zheng, Feng and Jin, Yaochu},
  journal={IEEE Transactions on Cybernetics},
  year={2024},
  publisher={IEEE}
}

@article{jiang2022survey,
  title={A survey of visual sensory anomaly detection},
  author={Jiang, Xi and Xie, Guoyang and Wang, Jinbao and Liu, Yong and Wang, Chengjie and Zheng, Feng and Jin, Yaochu},
  journal={arXiv preprint arXiv:2202.07006},
  year={2022}
}

Star History

Star History Chart