NWPU Campus
收藏资源简介:
NWPU Campus数据集是由西北工业大学计算机学院创建的大型半监督视频异常检测基准,包含43个场景、28类异常事件和16小时视频。该数据集是目前最大的半监督VAD数据集,也是首个考虑场景依赖异常和视频异常预测的数据集。数据集内容丰富,包括日常行为和多种异常行为,如单人异常、交互异常和场景依赖异常等。创建过程中,通过在校园户外地点设置摄像头记录行人和车辆活动,并由志愿者参与表演正常和异常事件。该数据集适用于视频异常检测和预测研究,旨在提前预警异常事件,以防止危险事故和减少生命财产损失。
The NWPU Campus Dataset is a large-scale semi-supervised video anomaly detection (VAD) benchmark developed by the School of Computer Science, Northwestern Polytechnical University. It includes 43 scenarios, 28 categories of anomalous events, and 16 hours of video footage. Currently, it is the largest semi-supervised VAD dataset, and also the first benchmark that considers scenario-dependent anomalies and video anomaly prediction. The dataset has rich content, covering both daily routine behaviors and various anomalous behaviors such as single-person anomalies, interactive anomalies, and scenario-dependent anomalies. During its construction, cameras were deployed at outdoor campus locations to record the activities of pedestrians and vehicles, and volunteers were invited to perform both normal and anomalous events. This dataset is applicable to research on video anomaly detection and prediction, aiming to provide early warnings for anomalous events to prevent dangerous accidents and reduce losses of lives and property.

- 1A New Comprehensive Benchmark for Semi-supervised Video Anomaly Detection and Anticipation西北工业大学计算机学院 · 2023年



