TrackEverything：用体素去重实现千帧级全点 3D 跟踪
TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations
做视频跟踪的可以看看，用体素去重加 3D WAFT，40GB 显存就能跟千帧视频的所有点，短片段比开源稠密跟踪器高 20% 以上。
TrackEverything 是一个 3D 点跟踪模型，把视频表示为世界坐标系下的持久 3D 场景轨迹，使模型复杂度与视频时长解耦。方法上采用滑动窗口边界的体素化去重机制合并共置轨迹，先用 endpoint refiner 预测终点和动静分类，再对动态点解码稠密轨迹，并用 3D WAFT 替代 4D 相关体积以降低显存开销。官方称这是首个能在 40 GB 显存内跟踪超过 1000 帧视频全部可见点的 3D 跟踪器。在 TAPVid-3D 基准上，短片段 APD 超越所有开源全帧稠密 3D 跟踪器 20% 以上，长序列上与稀疏跟踪器持平。
TrackEverything: Long Horizon Dense Tracking via De-Duplicating 3D Scene Representations
Existing point tracking models face a fundamental tradeoff: they can either track a sparse set of query points over long horizons, or track all points across only short clips. We introduce TrackEverything, a 3D point tracker that breaks this trade-off by representing videos as persistent 3D scene tracks in world coordinates. Grounded in the insight that videos are 2D projections of an underlying 3D world, TrackEverything decouples model complexity from video duration, allowing it to scale with unique physical scene geometry instead. Our approach introduces three key innovations. First, we employ a voxelization-based de-duplication mechanism at sliding-window boundaries to merge co-located tracks, preventing repeated observations of the same surface from redundantly accumulating. Second, we decompose tracking into an endpoint refiner that predicts each point's destination and static-versus-dynamic classification, followed by a lightweight trajectory refiner that decodes dense trajectories exclusively for dynamic points. Third, we propose 3D WAFT, replacing memory-prohibitive 4D correlation volumes with efficient feature sampling in the scene cloud. To the best of our knowledge, TrackEverything is the first 3D tracker capable of tracking all visible points across videos exceeding 1000 frames within 40 GB of GPU memory. On TAPVid-3D, TrackEverything outperforms all open-source all-frame dense 3D trackers by more than 20% APD on short clips, while remaining competitive with state-of-the-art sparse trackers on long sequences, despite tracking far more points.