LASER：面向免训练流式四维重建的逐层尺度对齐
LASER: Layer-wise Scale Alignment for Training-Free Streaming 4D Reconstruction
将离线前馈重建模型（如 VGGT、π³）转化为流式系统时，提出逐层尺度对齐替代全局 Sim(3) 对齐，解决单目尺度模糊导致的层间深度尺度不一致问题，可复用于任何需要跨时间窗口对齐深度预测的流式 3D 重建任务。
原文依据 we introduce layer-wise scale alignment, which segments depth predictions into discrete layers, computes per-layer scale factors
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期 VGGT 和 π³ 等前馈重建模型取得了令人印象深刻的重建质量，但由于二次内存复杂度而无法处理流式视频，限制了其实际部署。现有流式方法通过学习式记忆机制或因果注意力来解决该问题，但它们需要大量重新训练，且可能无法充分利用最优离线模型的强几何先验。我们提出 LASER，一个免训练框架，通过跨连续时序窗口对齐预测，将离线重建模型转换为流式系统。我们观察到，简单的相似变换（Sim(3)）对齐会因层深度错位而失败：单目尺度模糊导致不同场景层的相对深度尺度在窗口间不一致地变化。为此，我们引入逐层尺度对齐，将深度预测分割为离散层，计算每层尺度因子，并在相邻窗口和时间戳之间传播它们。大量实验表明，LASER 在相机姿态估计和点图重建上达到最优性能，同时在 RTX A6000 GPU 上以 14 FPS 和 6 GB 峰值内存运行，使公里级流式视频的实际部署成为可能。我们的代码将公开。
英文原文摘要 1353 字符
Recent feed-forward reconstruction models like VGGT and \pi^3 achieve impressive reconstruction quality but cannot process streaming videos due to quadratic memory complexity, limiting their practical deployment. While existing streaming methods address this through learned memory mechanisms or causal attention, they require extensive retraining and may not fully leverage the strong geometric priors of state-of-the-art offline models. We propose LASER, a training-free framework that converts an offline reconstruction model into a streaming system byaligning predictions across consecutive temporal windows. We observe that simple similarity transformation (Sim(3)) alignment fails due to layer depth misalignment: monocular scale ambiguity causes relative depth scales of different scene layers to vary inconsistently between windows. To address this, we introduce layer-wise scale alignment, which segments depth predictions into discrete layers, computes per-layer scale factors, and propagates them across both adjacent windows and timestamps.Extensive experiments show that LASER achieves state-of-the-art performance on camera pose estimation and point map reconstruction while operating at 14 FPS with 6 GB peak memory on a RTX A6000 GPU, enabling practical deployment for kilometer-scale streaming videos.Our code will be released publicly.