Dynamic Camera Poses and Where to Find Them
This addresses the critical need for scalable camera pose annotation in dynamic Internet videos to advance realistic video generation and simulation, representing a domain-specific incremental improvement.
The authors tackled the problem of annotating camera poses on dynamic Internet videos by introducing DynPose-100K, a large-scale dataset of 100,000 videos with camera poses, using a pipeline that combines task-specific and generalist models for filtering and improves pose estimation through point tracking, dynamic masking, and structure-from-motion techniques.
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.