CityFlow

CityFlow是一个城市规模的交通摄像机数据集，由来自10个十字路口的40个摄像机的超过3小时的同步高清视频组成，两个同时摄像机之间的最长距离正在2.5千米。该数据集包含超过200K个带注释的边界框，涵盖各种场景、视角、车辆模型和城市交通流量状况。
提供了相机的几何形状和校准信息，以帮助进行时空分析。此外，基准的子集可用于基于图像的车辆重新识别 (ReID) 的任务。

CityFlow is a city-scale traffic camera dataset, consisting of over 3 hours of synchronized high-definition video from 40 cameras deployed across 10 intersections. The maximum distance between any two simultaneously active cameras is 2.5 kilometers. This dataset contains over 200,000 annotated bounding boxes, covering diverse scenarios, viewing angles, vehicle models, and urban traffic flow conditions.

Camera geometry and calibration information are provided to facilitate spatio-temporal analysis. Additionally, a benchmark subset is available for image-based vehicle re-identification (ReID) tasks.