TrafficCAM
收藏资源简介:
TrafficCAM是由剑桥大学等机构创建的大规模交通流量图像数据集,包含4,402个带有语义和实例标注的图像帧以及59,944个未标注的图像帧。该数据集通过在印度八个城市的街道上安装的固定摄像头记录的视频序列构成,涵盖了多种车辆和行人。TrafficCAM旨在为全监督和半监督学习技术建立新基准,特别强调大量未标注数据的使用,以低成本的标注要求更好地捕捉交通流量质量。该数据集适用于交通流量分析,旨在优化交通管理,减少拥堵,并推动智能交通解决方案的发展。
TrafficCAM is a large-scale traffic flow image dataset developed by institutions including the University of Cambridge. It contains 4,402 image frames with semantic and instance annotations, as well as 59,944 unlabeled image frames. This dataset is constructed from video sequences recorded by fixed cameras installed on streets across eight cities in India, covering a variety of vehicle types and pedestrians. TrafficCAM aims to establish new benchmarks for fully supervised and semi-supervised learning technologies, with particular emphasis on utilizing large volumes of unlabeled data to better characterize traffic flow quality at a low annotation cost. This dataset is applicable to traffic flow analysis, with the goals of optimizing traffic management, reducing congestion, and promoting the development of intelligent transportation solutions.



