Realistic Anomaly Detection (RAD)
收藏资源简介:
Realistic Anomaly Detection (RAD) 数据集是由北京大学计算机科学学院等机构创建的,专门用于真实场景下的异常检测。该数据集包含4765张图像,涵盖13个类别和4种缺陷类型,通过机器人手臂从超过50个视角采集,提供了多视角的RGB数据。数据集的创建过程利用了机器人手臂的动态数据采集能力,确保了数据的多样性和真实性。RAD数据集主要应用于工业异常检测领域,旨在解决现有数据集在真实环境中的局限性,提升算法的鲁棒性和有效性。
Realistic Anomaly Detection (RAD) Dataset was developed by the School of Computer Science, Peking University and other institutions, specifically tailored for anomaly detection in real-world scenarios. This dataset consists of 4765 images spanning 13 categories and 4 defect types, and was collected via robotic arms from over 50 viewpoints, providing multi-view RGB data. The creation of the dataset leverages the dynamic data collection capability of robotic arms to ensure the diversity and authenticity of the collected data. The RAD dataset is primarily applied in the field of industrial anomaly detection, aiming to address the limitations of existing datasets in real-world environments and improve the robustness and effectiveness of related algorithms.

- 1RAD: A Dataset and Benchmark for Real-Life Anomaly Detection with Robotic Observations北京大学计算机科学学院, 迪肯大学信息技术学院, 清华大学人工智能产业研究院, 南京大学人工智能学院 · 2024年



