PoseTrackReID

PoseTrackReID是由马克斯普朗克计算机科学研究所创建的大规模数据集，旨在结合多人物姿态跟踪与视频人物重识别任务。该数据集包含46,933个标注帧，主要来源于PoseTrack 2018数据集的扩展，特别关注于高遮挡和运动模糊的挑战性场景。创建过程中，通过重新标注边界框和关键点，以及引入新的标注工具，确保了数据的准确性和完整性。PoseTrackReID的应用领域包括监控、体育分析和自动驾驶，旨在通过提供丰富的姿态信息和人物识别数据，解决复杂场景下的识别难题。

PoseTrackReID is a large-scale dataset created by the Max Planck Institute for Computer Science, designed to combine multi-person pose tracking and video person re-identification tasks. This dataset contains 46,933 annotated frames, which is mainly an extension of the PoseTrack 2018 dataset, with a particular focus on challenging scenarios with heavy occlusion and motion blur. During its creation, the accuracy and completeness of the data were ensured by re-annotating bounding boxes and key points, as well as introducing new annotation tools. The application fields of PoseTrackReID include surveillance, sports analytics, and autonomous driving, aiming to solve the recognition challenges in complex scenarios by providing rich pose information and person re-identification data.