← 研踪 PaperTrailarXiv:2611.22095 · v1
CueSplat: Learning to Schedule Sparse Deformation for Dynamic Gaussian Splatting on Edge Devices
首次提交 2026-11-17 10:24
版本更新 2026-11-17 10:24
雷达观测 2026-11-18 19:43
摘要
Deformation-based 4D Gaussian splatting (4DGS) enables dynamic scene rendering. However, evaluating deformation networks for all Gaussians at every frame is compute-intensive on edge devices. We find that Gaussian trajectories are often locally predictable, yet prediction reliability varies across Gaussians and time. Thus, we propose CueSplat, a post-training framework that exploits local temporal regularity while keeping the pretrained backbone frozen and retaining its Gaussian set. Specifically, CueSplat predicts Gaussian attributes from recent history and uses a lightweight learned router to select exact deformation updates within a scene-specific budget calibrated offline. Selection accounts for prediction reliability and rendering contribution. Selected indices are reused across nearby frames to reduce routing overhead, while selected Gaussians still receive exact updates each frame. To control accumulated prediction error, we use drift signals to schedule three consecutive full-deformation frames that rebuild the prediction history. Evaluated across four mainstream 4DGS algorithms, CueSplat achieves up to and speedups on Jetson AGX Orin and Orin NX, respectively, with a negligible quality drop (at most dB in dataset-average PSNR).
作者备注:10 pages, 6 figures. Main paper only
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-17 10:24 · 最新观测版本
CueSplat: Learning to Schedule Sparse Deformation for Dynamic Gaussian Splatting on Edge Devices
Deformation-based 4D Gaussian splatting (4DGS) enables dynamic scene rendering. However, evaluating deformation networks for all Gaussians at every frame is compute-intensive on edge devices. We find that Gaussian trajectories are often locally predictable, yet prediction reliability varies across Gaussians and time. Thus, we propose CueSplat, a post-training framework that exploits local temporal regularity while keeping the pretrained backbone frozen and retaining its Gaussian set. Specifically, CueSplat predicts Gaussian attributes from recent history and uses a lightweight learned router to select exact deformation updates within a scene-specific budget calibrated offline. Selection accounts for prediction reliability and rendering contribution. Selected indices are reused across nearby frames to reduce routing overhead, while selected Gaussians still receive exact updates each frame. To control accumulated prediction error, we use drift signals to schedule three consecutive full-deformation frames that rebuild the prediction history. Evaluated across four mainstream 4DGS algorithms, CueSplat achieves up to and speedups on Jetson AGX Orin and Orin NX, respectively, with a negligible quality drop (at most dB in dataset-average PSNR).
查看此版本解读 →查看此版本原文 ↗