AI Paper Atlas

Research map / Point clouds / Segmentation and detection

Weather-Robust LiDAR Perception: Point Cloud Restoration from Adverse Weather

Chenghao Sun, Pengpeng Sun, Xiangmo Zhao

AAAI 2026 · Point clouds › Segmentation and detection · approach: LiDAR voxel and pillar detectors · 1 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 spatial-frequency block (ablation)LTDNet (PE only) · IQA3D, PV-RCNNAP: 40.2 -> 42.4Improved
add probabilistic pruning block (ablation)LTDNet (PE+SFB) · IQA3D, PV-RCNNAP: 42.4 -> 43.1Improved
add weather-aware block (ablation)LTDNet (PE+SFB+PPB) · IQA3D, PV-RCNNAP: 43.1 -> 43.6Improved
preprocess degraded LiDAR with LTDNetPV-RCNN · RADIATE-W Rain (0-80m)AP@0.5: 38.43 -> 43.25Improved
preprocess degraded LiDAR with LTDNetPV-RCNN · RADIATE-W Snow (0-80m)AP@0.5: 36.53 -> 41.76Improved
preprocess with LTDNetPointPillars · RADIATE-W Fog (0-80m)AP@0.5: 29.19 -> 33.12Improved
preprocess with LTDNetCenterPoint · RADIATE-W Rain (0-80m)AP@0.5: 38.54 -> 42.87Improved
preprocess with denoising baselines (DSOR, 4DenoiseNet, 3D-UnOutDet)3D detectors · RADIATE-W/DENSE-WAP@0.5: typically less than 2 APNo clear effect
use LTDNet instead of denoising baselinespoint cloud quality · RADIATE/DENSE degraded scansDDI: baselines around 0.30; LTDNet 0.61 to 0.81Improved
denoising-only methods vs LTDNet on noise removalpoint cloud quality · RADIATE/DENSENR-F1: all about 90%No clear effect

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