Vehicles in the Middle East (VME)

VME数据集是一个专门为检测中东地区高分辨率卫星图像中的车辆而设计的数据集。该数据集由Maxar提供，涵盖了12个国家的54个城市，包含超过4000个图像瓦片和超过10万辆车辆，使用人工和半自动化方法进行注释。VME数据集旨在解决现有数据集在地理上的偏见问题，特别是对中东地区的忽视。数据集的创建过程涉及图像收集、图像裁剪、图像注释和最终数据集的形成。VME数据集和CDSI数据集的应用领域包括交通管理、城市规划、灾害响应等，旨在提高车辆检测的准确性和全球车辆检测的性能。

The VME dataset is a specialized collection purpose-built for vehicle detection in high-resolution satellite imagery across the Middle East. Provided by Maxar, the dataset covers 54 cities in 12 countries, includes over 4,000 image tiles and more than 100,000 annotated vehicles, with annotations generated via manual and semi-automated approaches. The VME dataset was developed to mitigate geographic bias in existing datasets, particularly the underrepresentation of the Middle East region. The dataset creation pipeline encompasses image collection, cropping, annotation, and final dataset curation. Applications of the VME and CDSI datasets span traffic management, urban planning, disaster response and other domains, with the goal of improving vehicle detection accuracy and overall global vehicle detection performance.