IPAD
收藏资源简介:
IPAD数据集是专为工业场景中的视频异常检测设计的,由上海交通大学创建。该数据集包含16种不同的工业设备,涵盖超过6小时的合成和真实世界视频素材,特别关注工业过程的关键特征——周期性。数据集的创建过程涉及现场工厂研究和与工程师的讨论,确保数据的真实性和代表性。IPAD数据集旨在解决工业生产中多种不可预测异常的检测问题,推动视频理解任务和智能工厂部署的发展。
The IPAD dataset, developed by Shanghai Jiao Tong University, is specifically designed for video anomaly detection in industrial scenarios. It contains 16 distinct types of industrial equipment, with over 6 hours of both synthetic and real-world video footage, with particular focus on periodicity—a key characteristic of industrial processes. The construction of the dataset involved on-site factory investigations and discussions with industrial engineers, ensuring the authenticity and representativeness of the collected data. The IPAD dataset aims to address the detection of various unpredictable anomalies in industrial production, and promote the advancement of video understanding tasks and the deployment of smart factories.

- 1IPAD: Industrial Process Anomaly Detection Dataset上海交通大学 · 2024年



