research
One Tracker Followed Every Visible Point Past 1,000 Frames
TrackEverything de-duplicated persistent 3D scene tracks to stay within 40 GB of GPU memory.
Summary
TrackEverything de-duplicated persistent 3D scene tracks to stay within 40 GB of GPU memory.
The system represents a video as persistent world-coordinate tracks, merges co-located observations at sliding-window boundaries and decodes dense trajectories only for points classified as dynamic. That design makes memory scale with unique scene geometry rather than raw video duration. The authors report more than 20% higher APD than open-source dense 3D trackers on short TAPVid-3D clips while remaining competitive with sparse trackers on longer sequences. These are benchmark results, not evidence of perfect tracking under every occlusion or camera motion.
Why it matters
TrackEverything de-duplicated persistent 3D scene tracks to stay within 40 GB of GPU memory.
Limits and context
- The system represents a video as persistent world-coordinate tracks, merges co-located observations at sliding-window boundaries and decodes dense trajectories only for points classified as dynamic.
- These are benchmark results, not evidence of perfect tracking under every occlusion or camera motion.
Key claims
TrackEverything de-duplicated persistent 3D scene tracks to stay within 40 GB of GPU memory.
Qualification: The system represents a video as persistent world-coordinate tracks, merges co-located observations at sliding-window boundaries and decodes dense trajectories only for points classified as dynamic.
Evidence: source-2026-09-26-006
Sources
- arXiv preprint 2609.30222arXiv · primary research
Corrections
No corrections have been recorded for this story.