AI Paper Atlas

Research map / Point clouds / Shape and surface completion

Cascaded Refinement Network for Point Cloud Completion

Xiaogang Wang, Marcelo H. Ang Jr., Gim Hee Lee

CVPR 2020 · Point clouds › Shape and surface completion · approach: Point cloud completion networks · 277 citations (OpenAlex)

Check 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
Use the cascaded refinement network with local patch adversarial lossPoint cloud completion · ShapeNet PCN test, AvgCD-P ($10^{-3}$): 9.636 -> 8.505Improved
Use the cascaded refinement networkPoint cloud completion · TopNet training data, resolutions 2048, 4096, 8192, 16384CD: 9.36 -> 7.61; 8.17 -> 6.57; 7.28 -> 5.72; 6.28 -> 5.21Improved
Use the cascaded refinement networkPoint cloud completion · Smaller training data, resolutions 2048, 4096, 8192, 16384CD ($10^{-4}$): 9.02 -> 7.57; 7.71 -> 6.71; 6.90 -> 5.84; 6.17 -> 5.21Improved
Use the cascaded refinement networkPoint cloud completion · Occlusion 20% to 70%CD ($10^{-4}$): 7.69 -> 5.52; 8.84 -> 6.72; 10.63 -> 8.46; 13.30 -> 11.36; 17.20 -> 15.26; 23.60 -> 21.27Improved
Use the cascaded refinement networkPoint cloud completion · Classification on completed test point cloudsAccuracy (%): 92.58 -> 94.17Improved
Add mean shape features (ablation)Cascaded refinement network · TopNet training data, 2048CD-P*: 7.78 -> 7.61Improved
Add the contraction-expansion unit (ablation)Cascaded refinement network · TopNet training data, 2048CD-P*: 7.83 -> 7.61Improved
Add the mirror operation for partial points (ablation)Cascaded refinement network · TopNet training data, 2048CD-P*: 7.67 -> 7.61Improved
Add the discriminator and adversarial training (ablation)Cascaded refinement network · TopNet training data, 2048CD-P*: 7.67 -> 7.61Improved

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