Search papers, labs, and topics across Lattice.
StreamCacheVGGT improves streaming 3D geometry reconstruction by addressing limitations in existing $O(1)$ memory frameworks that suffer from information loss due to binary token deletion and noisy single-layer scoring. The method introduces Cross-Layer Consistency-Enhanced Scoring (CLCES) to track token importance across transformer layers, and Hybrid Cache Compression (HCC) to merge moderately important tokens into retained anchors. Experiments across five benchmarks demonstrate state-of-the-art reconstruction accuracy and long-term stability under constant memory constraints.
Radically improve streaming 3D reconstruction by intelligently merging moderately important tokens instead of simply evicting them, leading to state-of-the-art accuracy.
Reconstructing dense 3D geometry from continuous video streams requires stable inference under a constant memory budget. Existing $O(1)$ frameworks primarily rely on a ``pure eviction''paradigm, which suffers from significant information destruction due to binary token deletion and evaluation noise from localized, single-layer scoring. To address these bottlenecks, we propose StreamCacheVGGT, a training-free framework that reimagines cache management through two synergistic modules: Cross-Layer Consistency-Enhanced Scoring (CLCES) and Hybrid Cache Compression (HCC). CLCES mitigates activation noise by tracking token importance trajectories across the Transformer hierarchy, employing order-statistical analysis to identify sustained geometric salience. Leveraging these robust scores, HCC transcends simple eviction by introducing a three-tier triage strategy that merges moderately important tokens into retained anchors via nearest-neighbor assignment on the key-vector manifold. This approach preserves essential geometric context that would otherwise be lost. Extensive evaluations on five benchmarks (7-Scenes, NRGBD, ETH3D, Bonn, and KITTI) demonstrate that StreamCacheVGGT sets a new state-of-the-art, delivering superior reconstruction accuracy and long-term stability while strictly adhering to constant-cost constraints.
