SMORE Adds Revisable External Memory to Frozen Geometry Models, Cuts Depth Error Up to 58%
zhenjun_zhao · x · 2026-10-21
SMORE equips frozen geometry foundation models with a revisable external memory for long-horizon inference, addressing error accumulation in windowed feed-forward models and the inability of streaming models to revise earlier estimates. Inspired by hippocampal associative memory, it stores, retrieves, verifies and revises past observations against new evidence. Across 8 geometry models and 5 datasets (pose estimation, multi-view depth, point-cloud reconstruction, SLAM, novel view synthesis), SMORE cuts multi-view depth error by up to 58% (e.g., DUST3R Replica overall error 12.02cm→3.55cm) and achieves 0.029m ATE RMSE on 7-Scenes long-horizon SLAM, beating many calibrated/uncalibrated SLAM baselines.
More from Research
- Laude Institute launches FrontierCS 2: agents must beat deleted paper contributions blind — lateinteraction · 2026-10-23
- RFD4-Proteina's 'underrated' docs detail partial diffusion for protein redesign — AllThingsApx · 2026-10-23
- MPI-Sec PhD Thesis on Vulnerability Detection Benchmarks Includes Pre-LLM Classic "Top Score on the Wrong Exam" — mboehme_ · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- New paper makes 9 falsifiable predictions toward a major AI-enabled terror attack by 2027-2028 — StephenLCasper · 2026-10-23
- HumanoidPose recovers humanoid robots' full-body joint states from RGB video alone — cholz · 2026-10-23