Research map / Detection and segmentation / Open-set and few-shot detection
Industrial anomaly detection
Detect anomalies with student-teacher, multi-sensor, and likelihood-based methods.
117 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Open-set and few-shot detection, keyword cluster “images · anomaly · interactive”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
34 papers in 2023–24 and 51 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 5.8% to 9.2%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Detection and Accurate Localization of Circular Fiducials Under Highly Challenging Conditions (CVPR 2016 · 88 citations)
- 2017 · IRINA: Iris Recognition (Even) in Inaccurately Segmented Data (CVPR 2017 · 34 citations)
- 2018 · Deep Anomaly Detection Using Geometric Transformations (NeurIPS 2018 · 398 citations)
- 2019 · MVTec AD -- A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection (CVPR 2019 · 2,068 citations)
- 2020 · Uninformed Students: Student-Teacher Anomaly Detection With Discriminative Latent Embeddings (CVPR 2020 · 914 citations)
- 2021 · Anomaly Detection in Video via Self-Supervised and Multi-Task Learning (CVPR 2021 · 355 citations)
- 2022 · Towards Total Recall in Industrial Anomaly Detection (CVPR 2022 · 1,653 citations)
- 2023 · DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection (CVPR 2023 · 280 citations)
- 2024 · Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detection (CVPR 2024 · 127 citations)
- 2025 · Forensics Adapter: Adapting CLIP for Generalizable Face Forgery Detection (CVPR 2025 · 46 citations)
- 2026 · RcAE: Recursive Reconstruction Framework for Unsupervised Industrial Anomaly Detection (AAAI 2026 · 1 citation)
Earliest papers here
- Detection and Accurate Localization of Circular Fiducials Under Highly Challenging Conditions (CVPR 2016 · 88 citations)
- iLab-20M: A Large-Scale Controlled Object Dataset to Investigate Deep Learning (CVPR 2016 · 61 citations)
- IRINA: Iris Recognition (Even) in Inaccurately Segmented Data (CVPR 2017 · 34 citations)
What has been tried
85 papers here report 898 measured changes (a component added, removed or swapped, compared with a baseline): helped 70%, no clear effect 5%, hurt 20%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add noisy anomalous training samples at α = 0.1 instead of α = 0.0 | Real-IAD · S-AUROC | 94.4 -> 92.2 | Hurt | Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detection (CVPR 2024 · 127 citations) |
| Add pixel-wise product aggregation of the 2D and 3D anomaly maps | MVTec 3D-AD · AUPRO@1% | 0.430 -> 0.455 | Helped | Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping (CVPR 2024 · 85 citations) |
| Remove masked blending boundary learning (L1) | CDF-v2 · AUC | 0.900 -> 0.899 | No clear effect | Forensics Adapter: Adapting CLIP for Generalizable Face Forgery Detection (CVPR 2025 · 46 citations) |
| Use Deepfake and Camera category prompts instead of Trump and Biden | UniversalFakeDetect · mAcc | 93.00 -> 93.79 | Helped | C2P-CLIP: Injecting Category Common Prompt in CLIP to Enhance Generalization in Deepfake Detection (AAAI 2025 · 43 citations) |
| Apply GFPGAN super-resolution to SRI | SRI · AUC | 0.772 -> 0.983 | Helped | DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · 42 citations) |
| Use proposed features and BRF with one segment | VAD high-shot supervised benchmark · Cl. AUROC | 87.3 -> 88.6 | Helped | Supervised Anomaly Detection for Complex Industrial Images (CVPR 2024 · 39 citations) |
| Add entropy loss (L_H) to supervised losses | MVTec LOCO AD, Average (LA) · AUROC | 95.9 -> 96.3 | Helped | Few Shot Part Segmentation Reveals Compositional Logic for Industrial Anomaly Detection (AAAI 2024 · 37 citations) |
| Replace region-to-region contrastive pretraining with SAM pretraining | Binary IFL test datasets · F1 fixed | 62.6 -> 48.6 | Hurt | SAFIRE: Segment Any Forged Image Region (AAAI 2025 · 15 citations) |
| Use RGB only instead of RGB, infrared, and point cloud | MulSen-AD · object-level AUROC | 0.961 -> 0.911 | Hurt | Multi-Sensor Object Anomaly Detection: Unifying Appearance, Geometry, and Internal Properties (CVPR 2025 · 11 citations) |
| Add the recovery branch | 3CAD · P-AUROC | +6.8% | Helped | 3CAD: A Large-Scale Real-World 3C Product Dataset for Unsupervised Anomaly Detection (AAAI 2025 · 10 citations) |
| include manually labeled anomalies in training data | FT-AED validation (5% FPR) · Miss Percentage | 25% -> 8.3% | Helped | FT-AED: Benchmark Dataset for Early Freeway Traffic Anomalous Event Detection (NeurIPS 2024 · 10 citations) |
| use ATSS as the cropped-image expert model instead of F-RCNN | LDXray cropped-image expert test · mAP | 48.9 -> 51.5 | Helped | Dual-view X-ray Detection: Can AI Detect Prohibited Items from Dual-view X-ray Images like Humans? (CVPR 2025 · 9 citations) |
What reviewers will expect
Datasets and baselines named in the 29 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- MVTec-AD · 2 of 29 papers
- CIFAR10 · 2 of 29 papers
- LVIS · 2 of 29 papers
- COCO · 2 of 29 papers
Baselines
- RD · 3 of 29 papers
- SimpleNet · 3 of 29 papers
- DeSTSeg · 2 of 29 papers
- UniAD · 2 of 29 papers
- CRAD · 2 of 29 papers
- F3Net · 2 of 29 papers
- UCF · 2 of 29 papers
- Faster R-CNN · 2 of 29 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 85 papers; from OpenAlex and the papers' affiliations).
- Tencent · 5 papers
- Tsinghua University · 5 papers
- Shanghai Jiao Tong University · 4 papers
- The Chinese University of Hong Kong · 3 papers
- Fudan University · 3 papers
- Tongji University · 3 papers
- Shanghai University · 2 papers
- Visual Intellgence · 2 papers
- Beijing Jiaotong University · 2 papers
- Vanderbilt University · 2 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Object-level representation learning (219 papers)
- Open-set and unknown object detection (160 papers)
- Interactive and referring segmentation (161 papers)
- Amodal and open-world instance segmentation (137 papers)
- Remote sensing segmentation and change (84 papers)
Elsewhere in the topic
- Unsupervised object and part discovery (114 papers)
- Domain-adaptive object detection (103 papers)
- Open-vocabulary object detection (96 papers)
- Open-vocabulary semantic segmentation (93 papers)
- Object-centric learning theory and video (86 papers)
- Few-shot object detection (67 papers)
- Human-object interaction detection (66 papers)
- Few-shot segmentation (64 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).