Event
TrackDeform3D: An RGB-D Deformable Object 3D Keypoint Tracking Framework and Dataset
1 reports · 1 sources
Yeheng Zong et al. propose TrackDeform3D, a markerless 3D keypoint tracking and dataset collection framework for deformable objects using only RGB-D cameras: it identifies 3D keypoints and tracks their trajectories combined with motion consistency constraints, generating temporally smooth and geometrically consistent data. The paper claims that, compared with multiple advanced tracking methods across several object categories, both geometric and tracking accuracy are improved. Using this framework, the authors released a dataset containing 6 deformable objects with a total of 110 minutes of trajectories. The paper has been accepted to IROS 2026.
Report timeline
- TrackDeform3D: A Markerless 3D Keypoint Tracking and Dataset Collection Framework for Deformable Objects Using RGB-D Cameras
TrackDeform3D is an affordable and autonomous framework for collecting 3D datasets of deformable objects using only RGB-D cameras. It identifies 3D keypoints and robustly tracks their trajectories with motion consistency constraints, producing temporally smooth and geometrically consistent data.
arXiv cs.RO ·