Awesome Industrial Anomaly Detection is a curated collection of papers and datasets for detecting anomalies and defects in industrial images using computer vision and deep learning.
The repository addresses the challenge of identifying manufacturing defects and anomalies in industrial inspection tasks. It organizes research literature and publicly available datasets relevant to this domain, spanning anomaly detection, anomaly segmentation, and defect detection methodologies. The collection is grounded in peer-reviewed survey work that examines the state of the field, including approaches ranging from normal-only training paradigms to those leveraging foundation model priors.
This resource suits researchers and practitioners working on industrial quality inspection, manufacturing defect detection, and related computer vision applications. It is particularly valuable for those seeking to understand the landscape of available datasets and published methods in industrial anomaly detection. The repository also connects to related projects including a benchmark for industrial image anomaly detection in manufacturing and a multimodal large language model specifically trained for industrial inspection tasks.
The project maintains active engagement with the research community through ongoing updates to its paper and dataset listings. The maintainers actively contribute new research directions, including recent work on anomaly synthesis methods, three-dimensional anomaly detection, and evaluation of multimodal models as industrial quality inspectors. The repository explicitly welcomes community contributions through pull requests for categorizing papers and adding relevant studies.
