AI Paper Atlas

Research map / 3D vision / 3D generation

Distilling Diffusion Models to Efficient 3D LiDAR Scene Completion

Shengyuan Zhang, An Zhao, Ling Yang, Zejian Li, Chenye Meng, Haoran Xu, Tianrun Chen, AnYang Wei, Perry Pengyun Gu, Lingyun Sun

ICCV 2025 · 3D vision › 3D generation · approach: Score distillation for text-to-3D · 7 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
add structural loss in distillation (ablation)ScoreLiDAR (Refined) · SemanticKITTICD: 0.419 -> 0.342Improved
add structural loss in distillation (ablation)ScoreLiDAR (Refined) · KITTI-360CD: 0.549 -> 0.452Improved
add scene-wise loss to structural loss (ablation)ScoreLiDAR · SemanticKITTICD: 0.367 -> 0.342Improved
add point-wise loss to structural loss (ablation)ScoreLiDAR · SemanticKITTICD: 0.351 -> 0.342Improved
lower scene weight to 0.05 from 0.5 (ablation)ScoreLiDAR · KITTI-360CD: 0.452 -> 0.494Hurt
add structural loss to LiDiff training (ablation)LiDiff (Refined) · SemanticKITTICD: 0.375 -> 0.399Hurt
distill LiDiff into ScoreLiDAR (5.37 s vs 30.55 s)LiDiff (Refined) · SemanticKITTICD: 0.375 -> 0.342Improved
use ScoreLiDAR distillation instead of fewer LiDiff steps (8 steps)LiDiff 8 steps Refined · SemanticKITTICD: 0.411 -> 0.342Improved
reduce sampling steps from 8 to 4 (ablation)ScoreLiDAR (Refined) · SemanticKITTICD: 0.342 -> 0.326Improved
reduce sampling steps from 8 to 1 (ablation)ScoreLiDAR (Refined) · SemanticKITTICD: 0.342 -> 0.548Hurt
use curvature-based key point selection instead of random (ablation)ScoreLiDAR · SemanticKITTICD: 0.384 -> 0.342Improved
use curvature-based key point selection instead of farthest (ablation)ScoreLiDAR · SemanticKITTICD: 0.442 -> 0.342Improved

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