CoTracker3
A video model focused on dense point tracking. It can predict long-term trajectories for arbitrary pixels in a video and maintains good continuity under occlusion or fast motion.
Timeline (last 90 days)
FastBench is constructed using a trajectory-driven data pipeline, comprising four steps: generating QA from high-frame-rate clips, filtering out questions answerable at low frame rate, using SAM3 and CoTracker3 for trajectory answer verification, and three rounds of manual inspection.
CoTracker3 is a model focused on dense point tracking, capable of long-term trajectory prediction for any pixel point in a video.
All Facts (2)
FastBench is constructed using a trajectory-driven data pipeline, comprising four steps: generating QA from high-frame-rate clips, filtering out questions answerable at low frame rate, using SAM3 and CoTracker3 for trajectory answer verification, and three rounds of manual inspection.
50%UnverifiedCoTracker3 is a model focused on dense point tracking, capable of long-term trajectory prediction for any pixel point in a video.
50%