Datasets › VisA

VisA (Visual Anomaly Dataset)

Introduced by Yang Zou et al. in SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation28 Jul 2022 archive 2025-07-28

The VisA dataset contains 12 subsets corresponding to 12 different objects as shown in the above figure. There are 10,821 images with 9,621 normal and 1,200 anomalous samples. Four subsets are different types of printed circuit boards (PCB) with relatively complex structures containing transistors, capacitors, chips, etc. For the case of multiple instances in a view, we collect four subsets: Capsules, Candles, Macaroni1 and Macaroni2. Instances in Capsules and Macaroni2 largely differ in locations and poses. Moreover, we collect four subsets including Cashew, Chewing gum, Fryum and Pipe fryum, where objects are roughly aligned. The anomalous images contain various flaws, including surface defects such as scratches, dents, color spots or crack, and structural defects like misplacement or missing parts.

Benchmarks archive 2025-07-28

All 3 leaderboards whose dataset resolves to this page shown (sort by any header). "First row" is the archive's own first row at snapshot, in the archive's row order; nothing here re-ranks and metric direction is not asserted.

First row (archive order)PaperCode
Anomaly Detection VisA UniNet Detection AUROC 99.8 UniNet: A Contrastive Learning-guided Unified Framework... pangdatangtt/UniNet 50 Compare
Anomaly Classification VisA APRIL-GAN Detection AUROC 78.0 APRIL-GAN: A Zero-/Few-Shot Anomaly Classification and... bychelsea/vand-april-gan +1 1 Compare
Multi-class Anomaly Detection VisA no rows — — 0 Compare

Papers archive 2025-07-28

30 shown of 41 papers with a leaderboard row on this dataset's benchmarks, newest first. The archive's own "papers using this dataset" list was never published, so this is the benchmark-backed subset; the archive's count for this dataset is 86. The Syntology column is from Syntology's graph (read 2026-09-24), stated per sample; it is not part of any archive number.

DateSamples run Syntology
Center-aware Residual Anomaly Synthesis for Multi-class Industrial Anomaly Detection 1 1 23 May 2025 not harvested
Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection 2 1 4 Mar 2025 ran 4 of 5 samples (1 unverified)
UniNet: A Contrastive Learning-guided Unified Framework with Feature Selection for Anomaly Detection 1 2 28 Feb 2025 not harvested
Progressive Boundary Guided Anomaly Synthesis for Industrial Anomaly Detection 1 1 23 Dec 2024 not harvested
Unlocking the Potential of Reverse Distillation for Anomaly Detection 1 1 10 Dec 2024 not harvested
SuperSimpleNet: Unifying Unsupervised and Supervised Learning for Fast and Reliable Surface Defect Detection 2 1 6 Aug 2024 not harvested
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection 1 1 22 Jul 2024 ran 19 of 28 samples (9 unverified)
VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation 1 1 17 Jul 2024 not harvested
A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localization 2 1 12 Jul 2024 ran 5 of 8 samples (3 unverified)
GLAD: Towards Better Reconstruction with Global and Local Adaptive Diffusion Models for Unsupervised Anomaly Detection 1 1 11 Jun 2024 ran 10 of 15 samples (5 unverified)
Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection 2 1 23 May 2024 ran 9 of 10 samples (1 unverified)
AnomalyDINO: Boosting Patch-based Few-shot Anomaly Detection with DINOv2 1 4 23 May 2024 ran 9 of 12 samples (3 unverified)
RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detection 1 1 9 Mar 2024 ran 9 of 10 samples (1 unverified)
MuSc: Zero-Shot Industrial Anomaly Classification and Segmentation with Mutual Scoring of the Unlabeled Images 1 1 30 Jan 2024 ran 1 of 1 samples (0 unverified)
Dynamic Addition of Noise in a Diffusion Model for Anomaly Detection 1 1 9 Jan 2024 not harvested
TransFusion -- A Transparency-Based Diffusion Model for Anomaly Detection 1 1 16 Nov 2023 ran 1 of 1 samples (0 unverified)
AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection 3 1 29 Oct 2023 ran 14 of 32 samples (18 unverified)
FAIR: Frequency-aware Image Restoration for Industrial Visual Anomaly Detection 1 1 13 Sep 2023 not harvested
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 2 2 27 May 2023 ran 4 of 10 samples (6 unverified)
Anomaly Detection with Conditioned Denoising Diffusion Models 1 1 25 May 2023 not harvested
Segment Any Anomaly without Training via Hybrid Prompt Regularization 2 1 18 May 2023 ran 3 of 9 samples (6 unverified)
Hard-normal Example-aware Template Mutual Matching for Industrial Anomaly Detection 1 1 28 Mar 2023 not harvested
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation 4 4 26 Mar 2023 ran 5 of 12 samples (7 unverified)
EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies 33 2 25 Mar 2023 ran 1 of 35 samples (34 unverified)
DiffusionAD: Norm-guided One-step Denoising Diffusion for Anomaly Detection 2 1 15 Mar 2023 not harvested
Reconstruction from edge image combined with color and gradient difference for industrial surface anomaly detection 1 1 26 Oct 2022 not harvested
Asymmetric Student-Teacher Networks for Industrial Anomaly Detection 1 1 14 Oct 2022 not harvested
DSR -- A dual subspace re-projection network for surface anomaly detection 1 1 2 Aug 2022 not harvested
SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation 1 1 28 Jul 2022 ran 0 of 2 samples (2 unverified)
AnoDDPM: Anomaly Detection With Denoising Diffusion Probabilistic Models Using Simplex Noise 1 1 30 Jun 2022 not harvested

The full list of 41 is in the JSON twin.

Dataset loaders archive 2025-07-28

1 loader as listed in the archive; links are outbound and not re-checked here.

Tasks archive 2025-07-28

License archive 2025-07-28

No licence recorded in the archive. Absence here is not a statement about the dataset's terms.

Modalities archive 2025-07-28

No modality tagged.

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

  • VisA

1 variant name, as the archive lists them.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections