Annotating camera poses on dynamic Internet videos at scale is critical for advancing fields like realistic video generation and simulation. However, collecting such a dataset is difficult, as most Internet videos are unsuitable for pose estimation. Furthermore, annotating dynamic Internet videos present significant challenges even for state-of-theart methods. In this paper, we introduce DynPose-100K, a large-scale dataset of dynamic Internet videos annotated with camera poses. Our collection pipeline addresses filtering using a carefully combined set of task-specific and generalist models. For pose estimation, we combine the latest techniques of point tracking, dynamic masking, and structure-from-motion to achieve improvements over the state-of-the-art approaches. Our analysis and experiments demonstrate that DynPose-100K is both large-scale and diverse across several key attributes, opening up avenues for advancements in various downstream applications.
@article{arxiv.2504.17788,
title = {Dynamic Camera Poses and Where to Find Them},
author = {Chris Rockwell and Joseph Tung and Tsung-Yi Lin and Ming-Yu Liu and David F. Fouhey and Chen-Hsuan Lin},
journal= {arXiv preprint arXiv:2504.17788},
year = {2025}
}
Comments
Accepted to CVPR 2025. Project Page: https://research.nvidia.com/labs/dir/dynpose-100k