MATCH is the first flow matching method for multi-view anomaly detection, reporting SOTA results on Real-IAD and the first comprehensive evaluation on MANTA-Tiny while enabling real-time use by omitting the divergence term.
arXiv preprint arXiv:2005.02357 , year=
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
fields
cs.CV 10representative citing papers
IDEAL learns intrinsic deviation vectors from normal and anomalous references via a Normal Variation Eraser and Intrinsic Deviation Encoder to score query deviations and generalize to unseen anomalies on eight datasets.
Align3D-AD improves zero-shot 3D anomaly detection by cross-modal feature alignment from RGB guidance and dual-prompt contrastive alignment to capture complementary semantics.
A training-free method fits PCA to DINOv2 features from few normal images and detects anomalies via reconstruction residual, reaching SOTA one-shot AUROC of 97.1% image-level on MVTec-AD and 93.2% on VisA.
Hypergraph model on DINOv2 tokens raises logical anomaly AUROC to 0.9279 on MVTec LOCO breakfast-box data by scoring an information quotient across local, relational, and hyperedge evidence.
ANoCo measures anomaly as the magnitude of the closed-form update that forces a query patch's features to satisfy normality constraints on a bipartite cosine-affinity graph with anchored normal nodes.
A new framework learns low-dimensional subspaces from nominal samples and reconstructs target deep embeddings via self-expressive linear combinations to localize anomalies, claiming SOTA on three benchmarks.
UniVAD v2 improves 1N-shot mean image-level AUC from 83.0% to 84.5% (85.7% with one abnormal reference) via support-conditioned boundary construction on six datasets.
MambaADv2 evolves Mamba state space models with hybrid blocks, frequency convolutions, and adaptive scanning for improved unsupervised anomaly detection.
Uni-RCM achieves state-of-the-art multi-class anomaly detection on MVTec-3D AD via a reference guide block and offline residual quantizer.
citing papers explorer
-
MATCH: Flow Matching for Multi-View Anomaly Detection
MATCH is the first flow matching method for multi-view anomaly detection, reporting SOTA results on Real-IAD and the first comprehensive evaluation on MANTA-Tiny while enabling real-time use by omitting the divergence term.
-
Beyond Normal References: Discriminative Few-Shot Anomaly Detection
IDEAL learns intrinsic deviation vectors from normal and anomalous references via a Normal Variation Eraser and Intrinsic Deviation Encoder to score query deviations and generalize to unseen anomalies on eight datasets.
-
Align3D-AD: Cross-Modal Feature Alignment and Dual-Prompt Learning for Zero-shot 3D Anomaly Detection
Align3D-AD improves zero-shot 3D anomaly detection by cross-modal feature alignment from RGB guidance and dual-prompt contrastive alignment to capture complementary semantics.
-
SubspaceAD: Training-Free Few-Shot Anomaly Detection via Subspace Modeling
A training-free method fits PCA to DINOv2 features from few normal images and detects anomalies via reconstruction residual, reaching SOTA one-shot AUROC of 97.1% image-level on MVTec-AD and 93.2% on VisA.
-
Hypergraph Normal World Models for Logical Visual Anomaly Detection
Hypergraph model on DINOv2 tokens raises logical anomaly AUROC to 0.9279 on MVTec LOCO breakfast-box data by scoring an information quotient across local, relational, and hyperedge evidence.
-
Anomaly as Non-Conformity via Training-Free Graph Laplacian Energy Minimization
ANoCo measures anomaly as the magnitude of the closed-form update that forces a query patch's features to satisfy normality constraints on a bipartite cosine-affinity graph with anchored normal nodes.
-
Subspace-Guided Feature Reconstruction for Unsupervised Anomaly Localization
A new framework learns low-dimensional subspaces from nominal samples and reconstructs target deep embeddings via self-expressive linear combinations to localize anomalies, claiming SOTA on three benchmarks.
-
UniVAD v2: Unified Visual Anomaly Detection via Support-Conditioned Boundary Construction
UniVAD v2 improves 1N-shot mean image-level AUC from 83.0% to 84.5% (85.7% with one abnormal reference) via support-conditioned boundary construction on six datasets.
-
MambaADv2: Evolving Duality-enhanced State Space Model for Unsupervised Anomaly Detection
MambaADv2 evolves Mamba state space models with hybrid blocks, frequency convolutions, and adaptive scanning for improved unsupervised anomaly detection.
-
Uni-RCM: Unified Reference-guided Cross-modal Mapping for Multi-Class Anomaly Detection
Uni-RCM achieves state-of-the-art multi-class anomaly detection on MVTec-3D AD via a reference guide block and offline residual quantizer.