Pith. sign in

REVIEW 15 cited by

FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flows

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2111.07677 v2 pith:XMDP36FS submitted 2021-11-15 cs.CV

Jiawei Yu , Ye Zheng , Xiang Wang , Wei Li , Yushuang Wu , Rui Zhao , Liwei Wu This is my paper · ORCID
classification cs.CV
keywords anomalydistributionfastflowdetectionfeaturefeaturesimageinference
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Unsupervised anomaly detection and localization is crucial to the practical application when collecting and labeling sufficient anomaly data is infeasible. Most existing representation-based approaches extract normal image features with a deep convolutional neural network and characterize the corresponding distribution through non-parametric distribution estimation methods. The anomaly score is calculated by measuring the distance between the feature of the test image and the estimated distribution. However, current methods can not effectively map image features to a tractable base distribution and ignore the relationship between local and global features which are important to identify anomalies. To this end, we propose FastFlow implemented with 2D normalizing flows and use it as the probability distribution estimator. Our FastFlow can be used as a plug-in module with arbitrary deep feature extractors such as ResNet and vision transformer for unsupervised anomaly detection and localization. In training phase, FastFlow learns to transform the input visual feature into a tractable distribution and obtains the likelihood to recognize anomalies in inference phase. Extensive experimental results on the MVTec AD dataset show that FastFlow surpasses previous state-of-the-art methods in terms of accuracy and inference efficiency with various backbone networks. Our approach achieves 99.4% AUC in anomaly detection with high inference efficiency.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 15 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SALAD -- Semantics-Aware Logical Anomaly Detection

    cs.CV 2025-09 conditional novelty 7.0 of 10

    SALAD trains a network on automatically extracted 'composition maps' of object parts and detects logical anomalies (missing/extra parts) with a 96.1% AUROC on MVTec LOCO, the best published result.

  2. SiM3D: Single-instance Multiview Multimodal and Multisetup 3D Anomaly Detection Benchmark

    cs.CV 2025-06 conditional novelty 7.0 of 10

    SiM3D provides a multiview, multimodal 3D anomaly detection benchmark with single-instance training and synthetic-to-real evaluation, showing adapted 2D methods often beat multimodal 3D methods on the new voxel-based task.

  3. DeCoFlow: Structural Decomposition of Normalizing Flows for Continual Anomaly Detection

    cs.CV 2026-06 unverdicted novelty 6.5 of 10

    DeCoFlow decomposes normalizing flow subnets into frozen bases and low-rank adapters with alignment, auxiliary layers, and tail-aware loss to achieve continual anomaly detection with zero forgetting and few added parameters.

  4. ReFP-AD: Rectified Flow Preconditioning for Energy-Based Anomaly Detection

    cs.LG 2026-08 conditional novelty 6.0 of 10

    ReFP-AD uses rectified-flow preconditioning to make finite-step MCMC stable for energy-based anomaly detection on full-dimensional DINOv2 tokens, achieving strong AUROC on MVTec-AD and VisA.

  5. ProCon: Projection-Consistency Memory for Training-Free Anomaly Detection

    cs.CV 2026-07 accept novelty 6.0 of 10

    Soft local projection residuals from seed- and depth-consensus normal memories outperform hard nearest-neighbor memory scoring for training-free industrial anomaly detection.

  6. TinyGLASS: Real-Time Self-Supervised In-Sensor Anomaly Detection

    cs.CV 2026-03 conditional novelty 6.0 of 10

    TinyGLASS runs self-supervised industrial anomaly detection directly on a Sony IMX500 sensor at 20 FPS with 8.6x parameter compression and 94.2% MVTec-AD image-level AUROC.

  7. PMODE: Theoretically Grounded and Modular Mixture Modeling

    cs.LG 2025-08 conditional novelty 6.0 of 10

    PMODE partitions data and fits per-subset density estimators, achieving near-optimal mixture rates for L1, L2, and KL objectives; its MV-PMODE variant scales to thousands of dimensions and rivals deep anomaly detectio...

  8. CLIP-Flow: A Universal Discriminator for AI-Generated Images Inspired by Anomaly Detection

    cs.CV 2025-08 conditional novelty 6.0 of 10

    A CLIP-feature normalizing-flow model trained on frequency-masked natural images, not on AI-generated images, detects images from unseen generators with 95.56% mAP on the UnivFD benchmark.

  9. An aerial color image anomaly dataset for search missions in complex forested terrain

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A new benchmark dataset of 34,424 labels across 10,659 aerial forest images from a real manhunt where standard anomaly detectors perform poorly.

  10. AutoDetect: Designing an Autoencoder-based Detection Method for Poisoning Attacks on Object Detection Applications in the Military Domain

    cs.CV 2025-09 conditional novelty 5.0 of 10

    AutoDetect, an autoencoder that flags images by the maximum reconstruction error across small slices, detects BadDet-style poisoned patches with AUROC 0.94+ on military and general object detection datasets.

  11. Toward Long-Tailed Online Anomaly Detection through Class-Agnostic Concepts

    cs.CV 2025-07 conditional novelty 5.0 of 10

    The paper proposes the long-tailed online anomaly detection (LTOAD) benchmark and a class-agnostic concept-based framework that outperforms class-aware baselines in most offline settings and in the online setting.

  12. Quantitative Benchmarking of Anomaly Detection Methods in Digital Pathology

    eess.IV 2025-06 conditional novelty 5.0 of 10

    A systematic comparison of 23 anomaly detection methods on pathology and industrial image datasets shows that feature distribution methods generally outperform reconstruction and distillation methods, and that epoch s...

  13. GCR: Geometry-Consistent Routing for Task-Agnostic Continual Anomaly Detection

    cs.CV 2026-01 conditional novelty 4.0 of 10

    GCR improves task-agnostic continual anomaly detection by routing in a shared frozen embedding space with geometry-consistent prototype matching, achieving near-zero forgetting on MVTec AD and VisA.

  14. NexViTAD: Few-shot Unsupervised Cross-Domain Defect Detection via Vision Foundation Models and Multi-Task Learning

    cs.CV 2025-07 reject novelty 4.0 of 10

    NexViTAD fuses Hiera and DINOv2 features with a multi-task decoder and Sinkhorn K-means memory bank to detect industrial defects across domains, reporting MVTec AD target AUC of 97.5%.

  15. Zero-Shot Anomaly Detection in Battery Thermal Images Using Visual Question Answering with Prior Knowledge

    cs.CV 2025-05 conditional novelty 4.0 of 10

    ChatGPT-4o, prompted with prior knowledge about normal battery temperature patterns, detects battery thermal anomalies at 86.6% AUC without training, though most trained methods score higher.

Pith tools