A curated list of research papers, datasets, and resources for anomaly detection in time-series, video, and image data.
Awesome Anomaly Detection is a curated GitHub repository listing academic papers, datasets, and resources for detecting anomalies—unusual patterns that deviate from expected behavior—in data. It addresses the need for a centralized, organized reference in a rapidly evolving research field, covering time-series, video, and image data across tasks like novelty detection, outlier detection, and one-class classification.
Machine learning researchers, data scientists, and engineers working on anomaly detection projects, particularly those exploring unsupervised or semi-supervised methods for industrial inspection, surveillance, fraud detection, or system monitoring.
It saves significant literature review time by aggregating and categorizing hundreds of papers from top venues, provides a clear taxonomy of methods, and includes links to code and datasets for practical implementation, making it a go-to starting point for both newcomers and experts.
A curated list of awesome anomaly detection resources
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Aggregates hundreds of papers from top conferences like CVPR and NeurIPS, organized by data type and detection task, saving significant literature review time.
Separates resources into dedicated sections for time-series, video-level, and image-level anomaly detection, with further subcategories for specific tasks like segmentation and OOD detection.
Many entries link to official implementations and benchmark datasets such as MVTec AD, facilitating practical experimentation and replication of methods.
Lists key survey papers from arXiv and other sources, providing foundational knowledge and tracking research trends, ideal for newcomers to the field.
Last updated in November 2021, missing recent advancements in a rapidly evolving field, which limits its usefulness for cutting-edge research.
The README explicitly states the time-series section 'need to survey more..', indicating gaps and potential imbalance in resource aggregation.
As a static GitHub repository, updates rely on infrequent community pull requests, with no recent activity suggesting limited ongoing curation or responsiveness.