Street Scene

该数据集包含了一系列视频，这些视频涉及空间环境异常情况，如闯红灯和不当车道使用等典型事件和各种空间环境异常。正常视频展示了行人走路和车辆/自行车在各自车道内正确行驶的场景。该数据集涵盖了17种不同的空间环境异常类型。规模上，它包含了46个训练视频和35个测试视频，总计有56,847个训练帧和146,410个测试帧。该数据集的任务是进行视频异常检测。

This dataset contains a series of videos involving spatial environment anomalies, including typical events such as running red lights and improper lane usage, as well as various other spatial environment anomalies. Normal videos showcase scenarios where pedestrians walk and vehicles/bicycles travel correctly within their respective lanes. This dataset covers 17 distinct types of spatial environment anomalies. In terms of scale, it includes 46 training videos and 35 test videos, with a total of 56,847 training frames and 146,410 test frames. The task of this dataset is video anomaly detection.