This repository contains the official implementation of the paper:

SubspaceAD: Training-Free Few-Shot Anomaly Detection via Subspace Modeling (CVPR2026)

Figure 1. SubspaceAD consists of two training-free stages: (1) extracting DINOv2 patch features from a few normal exemplars, and (2) estimating a low-dimensional PCA subspace to model normal appearance. Anomalies are detected via reconstruction residuals.

Detecting visual anomalies in industrial inspection often requires operating with only a few normal exemplars per category. While many recent approaches rely on large memory banks, auxiliary datasets, or multi-modal tuning, SubspaceAD introduces a minimalist, training-free alternative.

SubspaceAD consists of two stages:

1. Feature Extraction: Patch-level features are extracted from a small set of normal images using a frozen DINOv2 backbone.
2. Subspace Modeling: A PCA model is fit to these features to estimate the low-dimensional manifold of normal appearance.

At inference time, anomalies are detected using the reconstruction residual with respect to this learned subspace. Despite its simplicity, SubspaceAD achieves state-of-the-art performance in one-shot and few-shot settings.

Main Results (1-Shot)

• MVTec-AD: 98.0% Image AUROC; 97.6% Pixel AUROC
• VisA: 93.3% Image AUROC; 98.3% Pixel AUROC

MVTec-AD

VisA

Figure 2. Qualitative comparison on VisA and MVTec-AD (1-shot). SubspaceAD produces sharper and more precise anomaly maps than PromptAD and AnomalyDINO, with fewer false activations and better alignment with ground-truth defects across both datasets.

# 1. Create environment conda create -n subspacead python=3.10 conda activate subspacead # 2. Install dependencies and the package pip install -r requirements.txt pip install -e .

Download the dataset from the MVTec website and extract it to:

datasets/mvtec-ad/ 

# 1. Download and extract mkdir -p datasets/VisA_20220922 wget https://amazon-visual-anomaly.s3.us-west-2.amazonaws.com/VisA_20220922.tar tar -xvf VisA_20220922.tar -C datasets/VisA_20220922 # 2. Preprocess (reorganize folder structure) python tools/prepare_visa.py \ --data-folder datasets/VisA_20220922 \ --save-folder datasets/VisA_pytorch

pca-dino/ ├── datasets/ # Dataset root │ ├── mvtec-ad/ │ └── VisA_pytorch/ ├── logs/ # Experiment logs ├── scripts/ # SLURM/Bash benchmark scripts │ ├── benchmark_few_shot.sh │ ├── benchmark_full_shot.sh │ └── ... ├── src/ │ └── subspacead/ │ ├── core/ # PCA + feature extraction code │ ├── data/ # Dataset loaders & transforms │ ├── post_process/ # Scoring, pixel maps, filters │ └── utils/ # Visualization, logging ├── tools/ │ └── prepare_visa.py ├── main.py └── README.md 

Scripts are provided in scripts/ to reproduce all results. Edit MVTEC_PATH and VISA_PATH inside the scripts as needed.

Few-Shot (1, 2, 4 Shots) Selects k normal images per category.

bash scripts/benchmark_few_shot.sh

Batched Zero-Shot Fits PCA on the full unlabeled test set.

bash scripts/benchmark_batched0shot.sh

Full-Shot Uses all training images.

bash scripts/benchmark_full_shot.sh

Example: 1-shot MVTec-AD on bottle:

python main.py \ --dataset_name mvtec_ad \ --dataset_path datasets/mvtec-ad \ --categories bottle \ --model_ckpt facebook/dinov2-with-registers-giant \ --image_res 672 \ --k_shot 1 \ --aug_count 30 \ --pca_ev 0.99 \ --outdir results/debug_run

If you find this repository useful, please consider citing:

@misc{lendering2026subspaceadtrainingfreefewshotanomaly, title={SubspaceAD: Training-Free Few-Shot Anomaly Detection via Subspace Modeling}, author={Camile Lendering and Erkut Akdag and Egor Bondarev}, year={2026}, eprint={2602.23013}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2602.23013}, }