AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

FILT3R: Latent State Adaptive Kalman Filter for Streaming 3D Reconstruction

Seonghyun Jin, Jong Chul Ye

ECCV 2026 · Architectures and efficiency › Attention and transformers · approach: Block-sparse attention kernels · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace uniform overwrite/heuristic gate with Kalman-style adaptive filtering (FILT3R)CUT3R backbone · 7-Scenes (length 1000)Acc mean: 0.145 -> 0.054 (TTT3R vs FILT3R)Improved
replace TTT3R gating with FILT3RCUT3R backbone · 7-Scenes (length 300 and 500)Acc mean: 0.040 to 0.020 at length 300 and 0.066 to 0.024 at length 500Improved
replace TTT3R/CUT3R update with FILT3RCUT3R backbone · TUM-800ATE_orig: 0.214 -> 0.107Improved
replace TTT3R with FILT3RCUT3R backbone · TUM-800RPE-t: 0.009No clear effect
replace TTT3R with FILT3RCUT3R backbone · Bonn-300 (metric depth)Abs Rel: 0.108 to 0.093Improved
add periodic hard reset to TTT3R instead of FILT3RTTT3R · TUM-800ATE_orig: 0.107 vs 0.221Hurt
add drift-driven adaptive process noise (vs fixed Q) (ablation)FILT3R · TUM-800ATE_orig: 0.229 -> 0.107Improved
add drift-driven adaptive process noise (vs fixed Q) (ablation)FILT3R · Bonn-500 (metric depth)Abs Rel: 0.099 -> 0.089Improved
add variance propagation (vs reset-P) (ablation)FILT3R · TUM-800ATE_orig: 0.489 -> 0.107Improved
use Kalman gain instead of tuned fixed-beta EMA (0.05) (ablation)FILT3R · TUM-800ATE: 0.049 vs 0.057; RPE-r 0.658 -> 0.362Mixed
add EMA normalization of drift (ablation)FILT3R · TUM-800ATE_orig: 0.109 -> 0.107No clear effect
use fixed scalar measurement noise instead of adaptive r_t (ablation)FILT3R · TUM-800ATE_orig: 0.148 -> 0.107Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy