Anomaly Detection on VisA

Metric: Segmentation AUPRO (higher is better)

LeaderboardDataset

Loading chart...

Results

Hide augmentations

Sort:

#	Model↕	Segmentation AUPRO▼	Augmentations	Paper	Date↕	Code
1	DiffusionAD	96	No	DiffusionAD: Norm-guided One-step Denoising Diff...	2023-03-15	Code
2	Dinomaly ViT-L (model-unified multi-class)	94.8	No	Dinomaly: The Less Is More Philosophy in Multi-C...	2024-05-23	Code
3	INP-Former ViT-B (model-unified multi-class)	94.4	No	Exploring Intrinsic Normal Prototypes within a S...	2025-03-04	Code
4	GLAD	94.3	No	GLAD: Towards Better Reconstruction with Global ...	2024-06-11	Code
5	UniNet	93.9	No	-	-	Code
6	PBAS	93.3	No	Progressive Boundary Guided Anomaly Synthesis fo...	2024-12-23	Code
7	MuSc (zero-shot)	92.7	No	MuSc: Zero-Shot Industrial Anomaly Classificatio...	2024-01-30	Code
8	ReContrast	92.6	No	-	-	-
9	VCP-CLIP	90.7	Yes	VCP-CLIP: A visual context prompting model for z...	2024-07-17	Code
10	WinCLIP+ (4-shot)	87.6	No	WinCLIP: Zero-/Few-Shot Anomaly Classification a...	2023-03-26	Code
11	SuperSimpleNet	87.4	No	SuperSimpleNet: Unifying Unsupervised and Superv...	2024-08-06	Code
12	AnomalyCLIP	87	Yes	AnomalyCLIP: Object-agnostic Prompt Learning for...	2023-10-29	Code
13	APRIL-GAN	86.8	Yes	APRIL-GAN: A Zero-/Few-Shot Anomaly Classificati...	2023-05-27	Code
14	WinCLIP+ (2-shot)	86.2	No	WinCLIP: Zero-/Few-Shot Anomaly Classification a...	2023-03-26	Code
15	WinCLIP+ (1-shot)	85.1	No	WinCLIP: Zero-/Few-Shot Anomaly Classification a...	2023-03-26	Code
16	WinCLIP (0-shot)	56.8	No	WinCLIP: Zero-/Few-Shot Anomaly Classification a...	2023-03-26	Code

#1DiffusionADSOTA
96
Segmentation AUPRO· 2023-03-15
DiffusionAD: Norm-guided One-step Denoising Diffusion for Anomaly Detection Code
#2Dinomaly ViT-L (model-unified multi-class)
94.8
Segmentation AUPRO· 2024-05-23
Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection Code
#3INP-Former ViT-B (model-unified multi-class)
94.4
Segmentation AUPRO· 2025-03-04
Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection Code
#4GLAD
94.3
Segmentation AUPRO· 2024-06-11
GLAD: Towards Better Reconstruction with Global and Local Adaptive Diffusion Models for Unsupervised Anomaly Detection Code
#5UniNet
93.9
Segmentation AUPRO
No paperCode
#6PBAS
93.3
Segmentation AUPRO· 2024-12-23
Progressive Boundary Guided Anomaly Synthesis for Industrial Anomaly Detection Code
#7MuSc (zero-shot)
92.7
Segmentation AUPRO· 2024-01-30
MuSc: Zero-Shot Industrial Anomaly Classification and Segmentation with Mutual Scoring of the Unlabeled Images Code
#8ReContrast
92.6
Segmentation AUPRO
No paper
#9VCP-CLIP
90.7
Segmentation AUPRO· Augmentations· 2024-07-17
VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation Code
#10WinCLIP+ (4-shot)
87.6
Segmentation AUPRO· 2023-03-26
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation Code
#11SuperSimpleNet
87.4
Segmentation AUPRO· 2024-08-06
SuperSimpleNet: Unifying Unsupervised and Supervised Learning for Fast and Reliable Surface Defect Detection Code
#12AnomalyCLIP
87
Segmentation AUPRO· Augmentations· 2023-10-29
AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection Code
#13APRIL-GAN
86.8
Segmentation AUPRO· Augmentations· 2023-05-27
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 Code
#14WinCLIP+ (2-shot)
86.2
Segmentation AUPRO· 2023-03-26
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation Code
#15WinCLIP+ (1-shot)
85.1
Segmentation AUPRO· 2023-03-26
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation Code
#16WinCLIP (0-shot)
56.8
Segmentation AUPRO· 2023-03-26
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation Code