Research map / 3D vision / Face and shape
Combinative Matching for Geometric Shape Assembly
Nahyuk Lee, Juhong Min, Junhong Lee, Chunghyun Park, Minsu Cho
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 |
|---|---|---|---|
| use cosine affinity instead of L2 distance for occupancy (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.38 -> 0.28 | Improved |
| add orientation loss (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.31 -> 0.28 | Improved |
| add orientation loss (L2 affinity) (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.42 -> 0.38 | Improved |
| use equivariant backbone instead of DGCNN point embedding (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.74 -> 0.28 | Improved |
| use equivariant backbone instead of DGCNN point embedding (ablation) | CMNet · Breaking Bad everyday pairwise | RMSE(R): 38.74 -> 12.88 | Improved |
| add shape matching branch (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.38 -> 0.28 | Improved |
| add occupancy matching branch (ablation) | CMNet · Breaking Bad everyday pairwise | CRD: 0.35 -> 0.28 | Improved |
| add combinative (shape plus occupancy) matching vs prior state of the art PMTR | PMTR · Breaking Bad everyday pairwise | CRD: 0.39 -> 0.28 | Improved |
| use CMNet instead of PMTR | PMTR · Breaking Bad artifact pairwise | CRD: 0.60 -> 0.49 | Improved |
| use CMNet instead of PMTR in everyday-to-artifact transfer | PMTR · Breaking Bad transfer | CRD: 0.82 -> 0.74 | Improved |
| use CMNet instead of PMTR for multi-part assembly | PMTR · Breaking Bad multi-part everyday | PA_CRD: 66.95 -> 73.88 | Improved |
| use CMNet instead of Jigsaw for translation error | Jigsaw · Breaking Bad everyday pairwise | RMSE(T): 2.73 -> 3.78 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Jigsaw — Jigsaw: Learning to Assemble Multiple Fractured Objects (NeurIPS 2023) · better on 4 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- 3D Geometric Shape Assembly via Efficient Point Cloud Matching (ICML 2024 · similarity 0.90)
- Category-Level Multi-Part Multi-Joint 3D Shape Assembly (CVPR 2024 · similarity 0.89)
- Partial-to-Partial Shape Matching with Geometric Consistency (CVPR 2024 · similarity 0.87)
- Wormhole Loss for Partial Shape Matching (NeurIPS 2024 · similarity 0.85)
- Neural Assembler: Learning to Generate Fine-Grained Robotic Assembly Instructions from Multi-View Images (AAAI 2025 · similarity 0.85)
- Two by Two: Learning Multi-Task Pairwise Objects Assembly for Generalizable Robot Manipulation (CVPR 2025 · similarity 0.85)
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.
