Panda-Pose

Panda-Pose是由北卡罗来纳大学夏洛特分校开发的专为智能视频监控中的人体姿态估计设计的数据集。该数据集包含83000张训练图像和21000张验证图像，总计775000个标注的人体关键点。数据集通过高分辨率图像和用户定制的参数，如距离、拥挤度和遮挡密度，来模拟真实世界场景。Panda-Pose特别适用于解决远距离、高密度人群和严重遮挡条件下的人体姿态估计问题，显著提高了在复杂环境中的姿态估计准确性。Panda-Pose is a dataset developed by the University of North Carolina at Charlotte, specifically tailored for human pose estimation tasks in intelligent video surveillance systems. It comprises 83,000 training images and 21,000 validation images, with a total of 775,000 annotated human keypoints. The dataset simulates real-world scenarios by leveraging high-resolution images and user-defined parameters such as distance, crowd density, and occlusion density. Panda-Pose is particularly well-suited for addressing human pose estimation challenges under conditions of long-distance, high-density crowds and severe occlusion, thereby significantly improving the accuracy of pose estimation in complex environments.