AI Paper Atlas

Research map / Point clouds / Shape and surface completion

Point-Based Multi-View Stereo Network

Rui Chen, Songfang Han, Jing Xu, Hao Su

ICCV 2019 · Point clouds › Shape and surface completion · approach: Unsupervised shape correspondence · 417 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 edge convolution (relative-feature aggregation) (ablation)PointFlow · DTUaccuracy / completeness (mm): accuracy 0.501 -> 0.448; completeness 0.518 -> 0.487Improved
add Euclidean kNN grouping instead of pixel neighbours (ablation)PointFlow · DTUaccuracy / completeness (mm): accuracy 0.455 -> 0.448; completeness 0.492 -> 0.487Improved
add image feature pyramid (ablation)PointFlow · DTUaccuracy / completeness (mm): accuracy 0.455 -> 0.448; completeness 0.489 -> 0.487Improved
more flow iterations (0 to 3) (ablation)Point-MVSNet · DTUoverall (mm): 0.726 -> 0.391Improved
one flow iteration without upsampling vs coarse (ablation)Point-MVSNet · DTUoverall (mm): 0.726 -> 0.712Improved
increase point hypotheses from m=2 to m=1 (ablation)Point-MVSNet · DTUoverall (mm): 0.468 -> 0.479Hurt
increase point hypotheses from m=2 to m=3 (ablation)Point-MVSNet · DTUoverall (mm): 0.468 -> 0.484Hurt
more input views (2 to 5) (ablation)Point-MVSNet · DTUoverall (mm): 0.533 -> 0.468Improved
replace PU-Net upsampling with PointFlowcoarse point cloud · DTUoverall (mm): 0.943 -> 0.391Improved
apply PointFlow refinement to MVSNet depthMVSNet · Tanks and Temples intermediatef-score: 43.48 -> 48.27Improved
use pixel neighbours instead of Euclidean kNN (ablation)Point-MVSNet iteration 2 · DTUoverall (mm): 0.468 -> 0.474No clear effect
max pooling vs average pooling in edge convolution (ablation)PointFlow · DTUreconstruction errorNo 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