Research map / Image generation / Autoregressive and latent generation
HiFi-BRep: High-Fidelity Latent Representation for Robust B-Rep Generation
Junhao Hou, Chenqi Luo, Pufan Wang, Jiaying Lu, Yusheng Liu, Feiwei Qin, Meie Fang, Kun Zhou
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 one-stage joint decoding of geometry and topology (ablation) | HiFi-BRep · DeepCAD reconstruction | Valid (%): 69.3 -> 95.2 | Improved |
| use one-stage joint decoding of geometry and topology (ablation) | HiFi-BRep · DeepCAD reconstruction | Adj Acc (%): 73.2 -> 97.5 | Improved |
| add row-wise two-peak adjacency objective instead of independent BCE (ablation) | HiFi-BRep · DeepCAD reconstruction | Valid (%): 87.2 -> 95.2 | Improved |
| add row-wise two-peak adjacency objective (ablation) | HiFi-BRep · DeepCAD reconstruction | Adj Acc (%): 90.4 -> 97.5 | Improved |
| add encoder Topo-Mask (ablation) | HiFi-BRep · DeepCAD reconstruction | Valid (%): 89.5 -> 95.2 | Improved |
| add encoder Topo-Mask (ablation) | HiFi-BRep · DeepCAD reconstruction | Adj Acc (%): 92.7 -> 97.5 | Improved |
| add encoder Topo-Mask (ablation) | HiFi-BRep · DeepCAD reconstruction | Face Acc (%): 99.3 -> 100.0 | Improved |
| add canonicalization (lexicographic sorting of primitives) (ablation) | HiFi-BRep · DeepCAD reconstruction | Valid (%) | Improved |
| use HiFi-BRep single-stage design instead of DTGBrepGen | HiFi-BRep vs DTGBrepGen · DeepCAD unconditional generation | Valid (%): 43.20 -> 72.20 | Improved |
| use HiFi-BRep instead of DTGBrepGen | HiFi-BRep vs DTGBrepGen · ABC unconditional generation | Valid (%): 24.88 -> 32.66 | Improved |
| use HiFi-BRep instead of DTGBrepGen | HiFi-BRep vs DTGBrepGen · ABC unconditional generation | COV: 70.63 -> 57.93 | Hurt |
| use HiFi-BRep instead of DTGBrepGen | HiFi-BRep vs DTGBrepGen · DeepCAD unconditional generation | Compilability-Valid gap: 49.28 -> 18.18 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- DTGBrepGen — DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry (CVPR 2025) · better on 4 of 8 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
- BrepVGAE: Variational Graph Autoencoder with Unified Latent Representation for B-rep (CVPR 2026 · similarity 0.89)
- DTGBrepGen: A Novel B-rep Generative Model through Decoupling Topology and Geometry (CVPR 2025 · similarity 0.89)
- BRepNet: A Topological Message Passing System for Solid Models (CVPR 2021 · similarity 0.88)
- BrepGiff: Lightweight Generation of Complex B-rep with 3D GAT Diffusion (CVPR 2025 · similarity 0.88)
- Brep2Shape: Boundary and Shape Representation Alignment via Self-supervised Transformers (ICML 2026 · similarity 0.88)
- Flow-Corrected Shape Optimization: Taming Manifold Drift in High-Dimensional 3D Models (NeurIPS 2026 · similarity 0.83)
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.
