AI Paper Atlas

Research map / Point clouds / Shape and surface completion

Learning Normals of Noisy Points by Local Gradient-Aware Surface Filtering

Qing Li, Huifang Feng, Xun Gong, Yu-Shen Liu

ICCV 2025 · Point clouds › Shape and surface completion · approach: Point cloud upsampling and normals · 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 the extra loss terms (L_pd, L_sd, L_n, L_v) to the basic L_ld loss (ablation)LGSF · FamousShape unoriented normalsRMSE: 32.96 -> 19.98Improved
add L_n and L_v losses (ablation)LGSF · FamousShape unoriented normalsRMSE: 29.90 -> 20.00Improved
use all loss terms rather than subsets (ablation)LGSF · FamousShape oriented normalsRMSE: full 24.07 vs best subset 24.63Mixed
add neighborhood weighted normal aggregation at inference (ablation)LGSF · FamousShape unoriented normalsRMSE: 20.13 -> 19.98Improved
use neighborhood scale K=8 instead of 4 or 16 (ablation)LGSF · FamousShape unoriented normalsRMSE: 21.04 / 20.71 -> 19.98Improved
unsupervised LGSF compared with supervised methods on unoriented normalsPCPNet · PCPNet unoriented normals (average)RMSE: 12.25 vs DeepFit 11.80 and MSECNet 9.76Mixed
use LGSF instead of CAP-UDF under noiseunoriented normals · PCPNet (high noise)RMSE: 47.64 -> 22.49Improved
use LGSF instead of CAP-UDF on noise-free dataunoriented normals · PCPNet (no noise)RMSE: 7.59 -> 7.60No clear effect
unsupervised LGSF vs supervised denoisersPointCleanNet denoising · 10K 3% noiseCD (x10^4): 5.984 vs PD-Flow 5.956Mixed
add richer constraints versus NeuralGFconvergence · PCPNet oriented normalsRMSE at 10,000 iterations: 19.47 -> 19.22Improved
surface reconstruction on Lounge sceneLGSF · 3D Scene LoungeCD L2: 0.0221 (CAP-UDF) vs 0.1158Hurt

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