Skip to content

Research Reproduction · Python

fewshot-anomaly

Few-shot industrial anomaly detection: reproducing SubspaceAD (CVPR 2026) vs AnomalyDINO (WACV 2025) on DINOv2 features

What it is

The repository reproduces the CVPR 2026 SubspaceAD results for few-shot anomaly detection on the VisA dataset, providing an implementation of the PCA residual and nearest-neighbour scoring method, along with a FastAPI service and Gradio demo for defect inspection.

Scored at0df31f617 Nov 2026Changes since

Merit describes the code at this commit. Later pushes count once the owner rescans. How merit is scored

Distribution based on AI work

1 / 2
Machine Learning: 42.1%Computer Vision: 42.1%MLOps: 10.5%Data Science: 5.3%
  • Machine Learning42%
  • Computer Vision42%
  • MLOps11%
  • Data Science5%

Each field's share of the project's AI work, as scoring counts it. It ranks in the field it is most about.

What the judge found

Read from the project's code at its scored commit.

Strengths4

  • Successful reproduction of paper results within 0.3 AUROC points.
  • Includes a working demo and API for easy use and deployment.
  • Comprehensive test suite and CI ensure reliability and correctness.
  • Clear documentation of results, limitations, and reproduction steps.

Limitations

The owner keeps this project's limitations private.

Around it on the Machine Learning board

Whole board
  1. 49EmblaV1Architecture Modification / Research · Python57.9/100Implementationeze770
  2. 50LeNet_5-from-scratchArchitecture Reimplementation · Jupyter Notebook57.7/100ImplementationPop Alexandru
  3. 51, shared rankklinicka-knowledge-baseRAG System · HTML57.4/100ImplementationPetr Sovadina
  4. 51, shared rankmedical-mcq-generatorRAG System · Python57.4/100ImplementationMuhammad Iqbal Rahmatullah
  5. 51, shared rankfewshot-anomalyResearch Reproduction · Python57.4/100ImplementationRakif Khan Pathan

More from Rakif Khan Pathan