You can typically find the paper and related code on platforms like or the project's official GitHub repository by searching for "1pon dataset" or "Video-based 3D Person Reconstruction."
The file name 090622_001-1pon.mp4 follows the dataset's standard indexing, where 090622 typically refers to the capture date or session ID, and 001 identifies the specific subject or sequence. 090622_001-1pon.mp4
The dataset and paper were developed by researchers at Sensetime and the University of Tokyo . You can typically find the paper and related
The 1pon dataset specifically contains high-quality videos of approximately 100 individuals performing various actions, designed to test the robustness of human mesh recovery (HMR) algorithms. Key Aspects of the Dataset Key Aspects of the Dataset The paper establishes
The paper establishes a benchmark for reconstructing the 3D shape and pose of multiple people from monocular video sequences.