Research map / Point clouds / Shape and surface completion
Unpaired Point Cloud Completion on Real Scans using Adversarial Training
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use unpaired latent-space adversarial completion | Point-cloud completion · 3D-EPN chair | F1: 68.5 -> 78.0 | Helped |
| Use unpaired latent-space adversarial completion | Point-cloud completion · 3D-EPN chair | F1: 90.9 -> 78.0 | Hurt |
| Add adversarial training (ablation) | Ours point-cloud completion · 3D-EPN chair | F1: 54.4 -> 78.0 | Helped |
| Add the reconstruction loss term (ablation) | Ours point-cloud completion · 3D-EPN chair | F1: 75.3 -> 78.0 | Helped |
| Use the clean-complete AE latent space instead of an AE trained on partial inputs (ablation) | Ours point-cloud completion · 3D-EPN chair | F1: 65.4 -> 78.0 | Helped |
| Replace Hausdorff reconstruction loss with EMD (ablation) | Ours point-cloud completion · 3D-EPN chair | F1: 78.0 -> 77.5 | Hurt |
| Remove adversarial training from paired ground-truth training (ablation) | Ours+ · 3D-EPN car | F1: 79.2 -> 93.1 | Helped |
| Increase input incompleteness from 10 to 50 percent | Ours · Synthetic car | F1: 85.8 -> 72.5 | Hurt |
| Increase input incompleteness from 10 to 50 percent | Autoencoder baseline · Synthetic car | F1: 94.9 -> 63.4 | Hurt |
Closest papers by meaning
- GenPC: Zero-shot Point Cloud Completion via 3D Generative Priors (CVPR 2025 · similarity 0.88)
- Learning 3D Shape Completion From Laser Scan Data With Weak Supervision (CVPR 2018 · similarity 0.88)
- Diverse Shape Completion via Style Modulated Generative Adversarial Networks (NeurIPS 2023 · similarity 0.87)
- Unsupervised Point Cloud Completion and Segmentation by Generative Adversarial Autoencoding Network (NeurIPS 2022 · similarity 0.87)
- DC-PCN: Point Cloud Completion Network with Dual-Codebook Guided Quantization (AAAI 2025 · similarity 0.86)
- PointAttN: You Only Need Attention for Point Cloud Completion (AAAI 2024 · similarity 0.86)
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.