CityFlow
收藏资源简介:
CityFlow是一个城市规模的多目标多摄像头车辆跟踪和重识别基准数据集,由华盛顿大学和NVIDIA等机构创建。该数据集包含超过3小时的同步高清视频,来自10个交叉口的40个摄像头,覆盖了城市街道、住宅区和高速公路等多种场景。数据集包含超过20万个标注边界框,涵盖了广泛的场景、视角、车辆模型和城市交通流量条件。此外,数据集提供了摄像头几何和校准信息,以支持时空分析。CityFlow旨在推动城市交通优化研究,解决跨多个摄像头在不同天气条件下跟踪车辆的难题,并促进图像基础的车辆重识别技术的发展。
CityFlow is a city-scale multi-objective multi-camera vehicle tracking and re-identification benchmark dataset, developed by institutions including the University of Washington and NVIDIA. This dataset contains over 3 hours of synchronized high-definition videos captured by 40 cameras deployed at 10 intersections, covering diverse scenarios such as urban streets, residential areas, and highways. It contains more than 200,000 annotated bounding boxes, spanning a wide range of scenarios, viewing angles, vehicle models, and urban traffic flow conditions. In addition, the dataset provides camera geometry and calibration information to support spatiotemporal analysis. CityFlow aims to advance urban traffic optimization research, address the challenges of vehicle tracking across multiple cameras under varying weather conditions, and promote the development of image-based vehicle re-identification technologies.

- 1CityFlow: A City-Scale Benchmark for Multi-Target Multi-Camera Vehicle Tracking and Re-Identification华盛顿大学 · 2019年



