Research map / Point clouds / Shape and surface completion
Axolotl3D: a Unified Framework for Faithful 3D Shape Completion
Anita Hu, Maria Shugrina
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 changed | Where | Measured | Effect |
|---|---|---|---|
| train/test with single-view input, Axolotl3D vs ShapeR | Axolotl3D · Toys4k single-view with occlusion | F-score: 0.7453 -> 0.9046 | Improved |
| multi-view input, Axolotl3D vs ShapeR | Axolotl3D · OmniObject3D multi-view with occlusion | Chamfer distance (x10): 0.1443 (ShapeR) vs 0.1467 (Ours); F-score 0.9532 -> 0.9755 | Mixed |
| use ground-truth points vs predicted depth | Axolotl3D · Toys4k single-view | F-score: 0.9221 -> 0.7616 | Hurt |
| perturb camera parameters | Axolotl3D · Toys4k single-view predicted depth | F-score: 0.7616 -> 0.7620 | No clear effect |
| reduce input points to 2048 | ShapeR · Toys4k single-view with occlusion | F-score: 0.7453 -> 0.6956 | Hurt |
| add point cloud condition (ablation) | Axolotl3D · Toys4k occlusion | F-score | Improved |
| add mask-biased cross-attention (ablation) | Axolotl3D · Toys4k occlusion | F-score | Improved |
| use Plucker camera embeddings instead of learned per-view embeddings (ablation) | Axolotl3D · Toys4k single-view occlusion | F-score | Improved |
| skip ICP alignment at evaluation | all methods · Toys4k | F-score | Hurt |
Closest papers by meaning
- GENA3D: Generative Amodal 3D Modeling by Bridging 2D Priors and 3D Coherence (ECCV 2026 · similarity 0.89)
- Learning Generalizable Shape Completion with SIM(3) Equivariance (NeurIPS 2025 · similarity 0.88)
- RaySt3R: Predicting Novel Depth Maps for Zero-Shot Object Completion (NeurIPS 2025 · similarity 0.87)
- GeoComplete: Geometry-Aware Diffusion for Reference-Driven Image Completion (NeurIPS 2025 · similarity 0.87)
- PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025 · similarity 0.87)
- Sculpt3D: Multi-View Consistent Text-to-3D Generation with Sparse 3D Prior (CVPR 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.
