Research map / 3D vision / 3D generation
ROCA: Robust CAD Model Retrieval and Alignment From a Single Image
Check 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 end-to-end training with Procrustes alignment (ablation) | ROCA · ScanNet | class alignment accuracy: 14.9% to 19.4% | Helped |
| Add learned weights to Procrustes optimization (ablation) | ROCA P+E · ScanNet | class alignment accuracy: 19.4% to 20.4% | Helped |
| Add learned retrieval from predicted NOC correspondences (ablation) | ROCA P+E+W · ScanNet | class alignment accuracy: 20.4% to 21.5% | Helped |
| Use learned retrieval rather than Chamfer distance and add proxy completion (ablation) | ROCA · ScanNet | class retrieval-aware alignment accuracy: 15.9 -> 17.6 | Helped |
| Increase training images from 25k to 400k by denser video sampling (ablation) | ROCA · ScanNet | class alignment accuracy: 21.5 -> 24.4 | Helped |
| Add simulated Gaussian noise to NOC predictions, increasing σ from 0.0 to 0.3 (ablation) | ROCA · ScanNet | class alignment accuracy: 21.5 -> 19.2 | Hurt |
| Replace direct pose regression with dense correspondences and differentiable Procrustes alignment | ROCA · ScanNet | class retrieval-aware alignment accuracy: 9.5% to 17.6% | Helped |
Closest papers by meaning
- 3DCoMPaT200: Language Grounded Large-Scale 3D Vision Dataset for Compositional Recognition (NeurIPS 2024 · similarity 0.85)
- Tango3D: Towards Alignment for Global and Local 2D-3D Correspondence (NeurIPS 2026 · similarity 0.85)
- LASA: Instance Reconstruction from Real Scans using A Large-scale Aligned Shape Annotation Dataset (CVPR 2024 · similarity 0.85)
- CADCrafter: Generating Computer-Aided Design Models from Unconstrained Images (CVPR 2025 · similarity 0.85)
- Error Notebook-Guided, Training-Free Part Retrieval in 3D CAD Assemblies via Vision-Language Models (ICLR 2026 · similarity 0.85)
- From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach (AAAI 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.