DLOSAC refines RANSAC with Delaunay-guided spatial-geometric losses for geometry estimation
zhenjun_zhao · x · 2026-10-23
The paper proposes DLOSAC (Delaunay-Guided Local Optimization RANSAC) for multi-view geometry estimation:
- Pipeline: ORB feature matching → RANSAC inliers → Delaunay triangulation over inliers to capture neighborhood relations → spatial distribution loss plus edge-based geometric consistency loss → weighted least squares refinement
- Per-node edge and spatial losses are combined into weights via exp(-L/σ²)
- Across TUM, T&T, KITTI, CPC and EVD benchmarks, DLOSAC beats RANSAC, LO-RANSAC, GC-RANSAC and PSC-RANSAC on average error for fundamental matrix, relative pose and homography estimation, with near-zero inlier error on many metrics
More from Research
- Berkeley's StyleStream 2.0 cuts streaming voice conversion latency to 520ms, 2x faster — Berkeley · 2026-10-23
- Math PhD student quits academia: AI is making math unrecognizable — pickover · 2026-10-23
- ICLR Submission Found Containing LLM's Own Meta-Comment Admitting Results Weren't Significant — _onionesque · 2026-10-23
- AKV Serving System Lets Agents Manage Their Own KV Cache, Cutting Cost 27.7% — _akhaliq · 2026-10-23
- AuroraTTS-100K: a sub-100K-parameter English TTS running 1120x real-time on GPU — Tall_Abrocoma_3533 · 2026-10-23
- MiTaS multi-resolution tactile imitation learning hits 80% success vs 35.5% vision-only baseline — Jan_R_Peters · 2026-10-23