Eindhoven Marathon Dataset

Eindhoven Marathon Dataset是一个大规模的野外马拉松跑者视频数据集，由代尔夫特理工大学计算机视觉实验室创建。该数据集包含由42台手持智能手机相机录制的85小时高清视频，涵盖了9834名跑者。数据集的创建旨在解决在复杂场景中对跑者进行高效标注的挑战，特别是在多视角、光照变化、遮挡和人群密集等情况下。该数据集的应用领域包括计算机视觉中的对象检测、跟踪和人员重识别，特别是在非重叠视场的多相机设置中对跑者身份的精确对齐。The Eindhoven Marathon Dataset is a large-scale outdoor marathon runner video dataset created by the Computer Vision Laboratory of Delft University of Technology. This dataset contains 85 hours of high-definition video recorded by 42 handheld smartphone cameras, encompassing 9834 runners. The dataset was developed to address the challenges of efficiently annotating runners in complex scenarios, particularly under conditions including multi-view perspectives, varying illumination, occlusion, and dense crowds. Its applicable domains cover object detection, tracking, and person re-identification tasks in computer vision, especially the precise alignment of runner identities in multi-camera settings with non-overlapping fields of view.