CXR-AD
收藏资源简介:
CXR-AD数据集是首个针对半导体组件X射线图像的公开异常检测数据集,包含来自现实世界工业生产环境的真实X射线图像。数据集涵盖了五个工业组件类别,包括653个正常样本和561个缺陷样本,并具有精确的像素级掩码标注。该数据集旨在为异常检测算法提供新的视角,并为评估算法提供一个新基准。数据集提出了几个关键挑战,包括复杂背景和细微异常、多尺度异常分布以及高检测难度。
The CXR-AD dataset is the first publicly available anomaly detection dataset targeting X-ray images of semiconductor components. It contains real X-ray images collected from real-world industrial production environments. The dataset covers five industrial component categories, including 653 normal samples and 561 defective samples, and is equipped with precise pixel-level mask annotations. This dataset aims to provide a new perspective for anomaly detection algorithms and a novel benchmark for algorithm evaluation. Additionally, the dataset poses several key challenges, including complex backgrounds and subtle anomalies, multi-scale anomaly distributions, and high detection difficulty.




