发表机构
National Yang Ming Chiao Tung University; NVIDIA（国立阳明交通大学; 英伟达公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
Scal3R将在线三维重建转化为多参考相对位姿查询，通过轻量可学习标记和位姿图优化抑制漂移，在多数据集上实现性能提升
AI 中文摘要
在线三维重建模型在长视频上表现不佳，原因在于相对于固定的首帧锚点回归位姿会迫使模型外推至训练分布之外，微小漂移会累积并放大为严重的几何崩塌。然而，我们观察到每帧深度在整个失效过程中保持稳定，主干网络的局部几何结构完好无损，仅全局位姿头失效。受此解耦现象启发，我们提出Scal3R，该方法将在线重建重新表述为多参考相对位姿查询。我们使用约占参数总量1%的轻量可学习标记，通过非对称注意力机制注入完全冻结的主干网络，以此查询相对于多个过去关键帧的位姿。配备回环检测的在线位姿图优化系统可抑制长距离漂移。Scal3R在单个GPU上8小时即可收敛，在KITTI数据集上相较于在线基线将平均ATE降低超过60%，并在Virtual KITTI、Sintel、TUM-Dynamic、ScanNet和7-Scenes上达到了最先进的性能。项目页面：this https URL
英文摘要
Online 3D reconstruction models perform poorly on long videos. This happens because regressing poses relative to a fixed first-frame anchor forces extrapolation far beyond the training distribution. Small drifts accumulate and amplify into significant geometric collapse. However, we observe that per-frame depth remains stable throughout this failure. The backbone's local geometry remains intact; only the global pose head breaks down. Motivated by this decoupling, we introduce Scal3R. This approach reformulates online reconstruction as multi-reference relative pose querying. We use lightweight learnable tokens, which make up about ~1% of the parameters, and inject them into a completely frozen backbone via asymmetric attention. This setup queries poses relative to multiple past keyframes. An online pose-graph optimization system with loop closure suppresses long-range drift. Scal3R reaches convergence in 8 hours on a single GPU. It reduces the average ATE by over 60% on KITTI compared to the online baseline. It also achieves state-of-the-art performance across Virtual KITTI, Sintel, TUM-Dynamic, ScanNet, and 7-Scenes. Project page: https://linjohnss.github.io/scal3r/
CommentsECCV 2026. Project page: https://linjohnss.github.io/scal3r/
