Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction
2026-04-09 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors focus on building detailed 3D models of large scenes from long videos. They note that existing methods struggle to keep accurate and consistent results over long sequences because they can't remember enough about the whole scene. To fix this, the authors created a new way to store and use global scene information efficiently, which helps improve 3D reconstruction accuracy and stability. Their approach uses small neural networks that learn during testing without much extra computation. They tested their method on big datasets and showed it performs better in accuracy while staying efficient.
3D scene reconstructionRGB imagesglobal context representationneural networksself-supervised learningpose accuracylarge-scale datasetsKITTI OdometryOxford Spires
Authors
Tao Xie, Peishan Yang, Yudong Jin, Yingfeng Cai, Wei Yin, Weiqiang Ren, Qian Zhang, Wei Hua, Sida Peng, Xiaoyang Guo, Xiaowei Zhou
Abstract
This paper addresses the task of large-scale 3D scene reconstruction from long video sequences. Recent feed-forward reconstruction models have shown promising results by directly regressing 3D geometry from RGB images without explicit 3D priors or geometric constraints. However, these methods often struggle to maintain reconstruction accuracy and consistency over long sequences due to limited memory capacity and the inability to effectively capture global contextual cues. In contrast, humans can naturally exploit the global understanding of the scene to inform local perception. Motivated by this, we propose a novel neural global context representation that efficiently compresses and retains long-range scene information, enabling the model to leverage extensive contextual cues for enhanced reconstruction accuracy and consistency. The context representation is realized through a set of lightweight neural sub-networks that are rapidly adapted during test time via self-supervised objectives, which substantially increases memory capacity without incurring significant computational overhead. The experiments on multiple large-scale benchmarks, including the KITTI Odometry~\cite{Geiger2012CVPR} and Oxford Spires~\cite{tao2025spires} datasets, demonstrate the effectiveness of our approach in handling ultra-large scenes, achieving leading pose accuracy and state-of-the-art 3D reconstruction accuracy while maintaining efficiency. Code is available at https://zju3dv.github.io/scal3r.