Anomaly Detection on MVTec AD

157개 결과 · ⬇ CSV · JSON

Detection AUROC

Rank	Model	Detection AUROC	Segmentation AUPRO	Segmentation AUROC	Segmentation AP	FPS	Extra Training Data	Paper	Code	Year
1	CAD 자동 추출	100	–	–	–	–		CADIC: Continual Anomaly Detection Based on Incremental Coreset		2025
2	GLASS	99.9	96.8	99.3	–	–	✓	A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localization	cqylunlun/glass · septmars/DL	2024
2	UniNet	99.90	96.00	98.81	–	–		UniNet: A Contrastive Learning-guided Unified Framework with Feature Selection for Anomaly Detection	pangdatangtt/UniNet	2025
4	PBAS	99.8	97.3	98.6	–	–		Progressive Boundary Guided Anomaly Synthesis for Industrial Anomaly Detection	cqylunlun/pbas	2024
4	HETMM	99.8	96.4	99	–	–		Hard-normal Example-aware Template Mutual Matching for Industrial Anomaly Detection	NarcissusEx/HETMM	2023
4	INP-Fomer ViT-L (model-unified multi-class)	99.8	95.6	98.6	72.1	–		Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection	luow23/inp-former · septmars/DL	2025
4	DDAD	99.8	–	98.1	–	–	✓	Anomaly Detection with Conditioned Denoising Diffusion Models	arimousa/DDAD	2023
4	EfficientAD (early stopping)	99.8	–	–	–	269	✓	EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies	openvinotoolkit/anomalib · nelson1425/EfficientAD · rximg/EfficientAD · +30	2023
9	Dinomaly ViT-L (model-unified multi-class)	99.77	95.09	98.54	70.53	–	✓	Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection	guojiajeremy/dinomaly · septmars/DL	2024
10	ReConPatch Ensemble (+RefineNet)	99.72	–	99.2	–	–		ReConPatch : Contrastive Patch Representation Learning for Industrial Anomaly Detection	travishsu/ReConPatch-TF	2023
11	ReConPatch WRN-50 (+RefineNet)	99.71	–	98.62	–	–		ReConPatch : Contrastive Patch Representation Learning for Industrial Anomaly Detection	travishsu/ReConPatch-TF	2023
12	ADClick	99.7	97.8	99.2	82.9	–	✓	Towards Efficient Pixel Labeling for Industrial Anomaly Detection and Localization		2024
12	CPR	99.7	97.8	99.2	82.7	113	✓	Target before Shooting: Accurate Anomaly Detection and Localization under One Millisecond via Cascade Patch Retrieval	flyinghu123/cpr	2023
12	CPR-fast	99.7	97.7	99.2	82.3	245	✓	Target before Shooting: Accurate Anomaly Detection and Localization under One Millisecond via Cascade Patch Retrieval	flyinghu123/cpr	2023
12	MSFlow	99.7	97.1	98.8	–	–		MSFlow: Multi-Scale Flow-based Framework for Unsupervised Anomaly Detection	cool-xuan/msflow	2023
12	CRAS	99.7	–	98.4	–	–		Center-aware Residual Anomaly Synthesis for Multi-class Industrial Anomaly Detection	cqylunlun/CRAS	2025
17	PNI Ensemble	99.63	96.55	99.06	–	–		PNI : Industrial Anomaly Detection using Position and Neighborhood Information	wogur110/PNI_anomaly_detection	2022
18	ReConPatch WRN-101	99.62	–	98.53	–	–		ReConPatch : Contrastive Patch Representation Learning for Industrial Anomaly Detection	travishsu/ReConPatch-TF	2023
19	WeakREST-Un	99.6	97.6	99.3	83.0	25.2		Industrial Anomaly Detection and Localization Using Weakly-Supervised Residual Transformers		2023
19	Dinomaly ViT-B (model-unified multi-class)	99.60	94.79	98.35	69.29	–	✓	Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection	guojiajeremy/dinomaly · septmars/DL	2024