Research map / 3D vision / Face and shape
DeepMend: Learning Occupancy Functions to Represent Shape for Repair
Nikolas Lamb, Sean Banerjee, Natasha Kholgade Banerjee
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 |
|---|---|---|---|
| add both mean-occupancy and mean-proximity penalties (L_infaug) to inference (ablation) | DeepMend inference · ShapeNet | NE%: 91.5% -> 97.6% | Improved |
| add mean proximity penalty L_prox alongside the non-empty term (ablation) | DeepMend inference · ShapeNet | NE%: 80.1% -> 97.6% | Improved |
| add per-point non-empty penalty L_nerp (ablation) | DeepMend inference · ShapeNet | Chamfer distance: 0.096 -> 0.275 | Hurt |
| add per-point non-empty penalty L_nerp (ablation) | DeepMend inference · ShapeNet | NE%: 91.5% -> 99.7% | Improved |
| add mean non-empty penalty L_ner only (ablation) | DeepMend inference · ShapeNet | Chamfer distance: 0.096 -> 0.159 | Hurt |
| add per-point proximity penalty L_proxp (ablation) | DeepMend inference · ShapeNet | Chamfer distance: 0.096 -> 0.055 | Improved |
| add per-point proximity penalty L_proxp (ablation) | DeepMend inference · ShapeNet | NE%: 91.5% -> 73.4% | Hurt |
| use DeepMend instead of 3D-ORGAN | DeepMend · ShapeNet | Chamfer distance: 0.250 -> 0.063 | Improved |
| use DeepMend instead of Sub-Occ | DeepMend · ShapeNet | NFRE: 0.128 -> 0.011 | Improved |
| use DeepMend instead of Sub-Lamb | DeepMend · ShapeNet | NFRE: 0.250 -> 0.011 | Improved |
| use DeepMend instead of Sub-Occ | DeepMend · ShapeNet | Chamfer distance: 0.067 -> 0.063 | Mixed |
Closest papers by meaning
- Beyond Reassembly: Fractured Object Recovery with Missing Parts (CVPR 2026 · similarity 0.87)
- UniRestore3D: A Scalable Framework For General Shape Restoration (ICLR 2025 · similarity 0.87)
- Fantastic Breaks: A Dataset of Paired 3D Scans of Real-World Broken Objects and Their Complete Counterparts (CVPR 2023 · similarity 0.87)
- Breaking Bad: A Dataset for Geometric Fracture and Reassembly (NeurIPS 2022 · similarity 0.85)
- GARF: Learning Generalizable 3D Reassembly for Real-World Fractures (ICCV 2025 · similarity 0.85)
- Learning Partonomic 3D Reconstruction from Image Collections (CVPR 2025 · similarity 0.84)
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.
