LoGeR: Long-Context Geometric Reconstruction with Hybrid Memory
This work addresses the problem of long-context geometric reconstruction for applications requiring dense 3D video understanding, such as autonomous driving or robotics, with an incremental yet significant improvement over prior state-of-the-art feedforward methods.
The authors tackled the problem of scaling dense 3D reconstruction to long video sequences, achieving a 74% reduction in ATE on the KITTI dataset and enabling reconstruction over thousands of frames. LoGeR generalizes to sequences of up to 19k frames during inference.
Feedforward geometric foundation models achieve strong short-window reconstruction, yet scaling them to minutes-long videos is bottlenecked by quadratic attention complexity or limited effective memory in recurrent designs. We present LoGeR (Long-context Geometric Reconstruction), a novel architecture that scales dense 3D reconstruction to extremely long sequences without post-optimization. LoGeR processes video streams in chunks, leveraging strong bidirectional priors for high-fidelity intra-chunk reasoning. To manage the critical challenge of coherence across chunk boundaries, we propose a learning-based hybrid memory module. This dual-component system combines a parametric Test-Time Training (TTT) memory to anchor the global coordinate frame and prevent scale drift, alongside a non-parametric Sliding Window Attention (SWA) mechanism to preserve uncompressed context for high-precision adjacent alignment. Remarkably, this memory architecture enables LoGeR to be trained on sequences of 128 frames, and generalize up to thousands of frames during inference. Evaluated across standard benchmarks and a newly repurposed VBR dataset with sequences of up to 19k frames, LoGeR substantially outperforms prior state-of-the-art feedforward methods--reducing ATE on KITTI by over 74%--and achieves robust, globally consistent reconstruction over unprecedented horizons.