Anomaly Detection On Mvtec Ad

Metriken

Detection AUROC

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	Detection AUROC	Paper Title	Repository
CutPaste+SSPCAB	96.1	Self-Supervised Predictive Convolutional Attentive Block for Anomaly Detection
EfficientAD-M	99.1	EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies
PatchCore Large	99.6	Towards Total Recall in Industrial Anomaly Detection
CPR-faster	99.4	Target before Shooting: Accurate Anomaly Detection and Localization under One Millisecond via Cascade Patch Retrieval
Fastflow	99.4	FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flows
CPR-fast(TensorRT)	-	Target before Shooting: Accurate Anomaly Detection and Localization under One Millisecond via Cascade Patch Retrieval
WinCLIP+ (1-shot)	93.1	WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation
DSR	98.2	DSR -- A dual subspace re-projection network for surface anomaly detection
CAVGA-D (weakly-supervised)	-	Attention Guided Anomaly Localization in Images	-
Gaussian-AD+DFS	96.6	Deep Feature Selection for Anomaly Detection Based on Pretrained Network and Gaussian Discriminative Analysis
ReConPatch Ensemble	-	ReConPatch : Contrastive Patch Representation Learning for Industrial Anomaly Detection
EfficientAD-S	98.7	EfficientAD: Accurate Visual Anomaly Detection at Millisecond-Level Latencies
FYD	97.7	Focus Your Distribution: Coarse-to-Fine Non-Contrastive Learning for Anomaly Detection and Localization	-
Student–Teacher AD (p=65)	-	Uninformed Students: Student-Teacher Anomaly Detection with Discriminative Latent Embeddings
ProbabilisticPatchCore	98.2	A Probabilistic Transformation of Distance-Based Outliers
UTAD	90	Unsupervised Two-Stage Anomaly Detection	-
GLASS	99.9	A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localization
MemSeg	99.56	MemSeg: A semi-supervised method for image surface defect detection using differences and commonalities
CutPaste (Patch level detector)	-	CutPaste: Self-Supervised Learning for Anomaly Detection and Localization
THFR	99.2	Template-guided Hierarchical Feature Restoration for Anomaly Detection	-

0 of 144 row(s) selected.