This approach reformulates online reconstruction as multi-reference relative pose querying, which achieves state-of-the-art performance across Virtual KITTI, Sintel, TUM-Dynamic, ScanNet, and 7-Scenes and reaches convergence in 8 hours on a single GPU.
Abstract
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/
Feedforward foundation models have recently shown remarkable 3D reconstruction capabilities. However, existing models exhibit large tracking drift in long-context streaming reconstruction due to error accumulation. In this paper, we revisit loop closure with streaming reconstruction foundation models to enable accurate...
Mo-Yang Li, Zihan Zhu, Wei Zhang et al.· 0 citations
This work proposes LoG-VGGT, a memory-efficient framework for long-sequence 3D reconstruction that balances local temporal modeling with global camera consistency, and introduces cross-window attention at a small subset of transformer blocks, enabling effective information propagation across adjacent temporal windows w...
Jing-Ke Zhou, Chen-Hang Ma, Zhi-Zhou Zhong et al.· 1 citation
Pretrained 3D encoders are typically developed on globally reconstructed scenes expressed in a consistent world coordinate frame, whereas embodied systems must reason from partial, viewpoint-dependent observations in camera coordinates. We show that this shift from globally learned 3D feature spaces to realistic partia...
Akira-Miranda Adeyomi Adeniran-Lowe, B. Singh, Lars Arnold Dethlefsen et al.· 0 citations
ABot-Recon is presented, a simple streaming model that caches KV features from only the preceding 11 frames and predicts a point map in the current camera coordinate system together with an adjacent-frame relative pose that remains equivariant under changes of reference frame.
Jiarong Han, Jincheng Xiong, Yu-Zhou Liu et al.· 2 citations· ⚡1
This work proposes Scene Geometric Invariant Anchoring (SGIA), which extracts dominant geometric invariants from each chunk's predicted point cloud via coarse-to-fine robust estimation and exploits their cross-chunk consistency to establish scale constraints independent of point cloud registration.
Wei Zhang, Yihang Wu, Song Li et al.· 2 citations· ⚡1
This work presents a deep, global Structure-from-Motion framework based on learned view-graph aggregation that employs a permutation-equivariant, edge-conditioned graph neural network that takes noisy pairwise relative poses as input and outputs globally consistent camera extrinsics.
Fadi Khatib, M. Galun, R. Basri· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.