Research map / 3D vision / Face and shape
Neural Compression for 3D Geometry Sets
Siyu Ren, Junhui Hou, Weiyao Lin, Wenping Wang
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 TSDF-Def with deformable marching cubes instead of TSDF (ablation) | regular representation · Mixed (thin structures) | CD (x1e-3): 6.630 -> 5.396 | Improved |
| use TSDF-Def instead of TSDF (ablation) | regular representation · Mixed (thin structures) | F1-0.005: 0.621 -> 0.677 | Improved |
| add L1 regularization on deformation (ablation) | TSDF-Def · Mixed (thin structures) | CD (x1e-3): 6.205 -> 5.396 | Improved |
| add L1 regularization on deformation (ablation) | TSDF-Def · Mixed (thin structures) | F1-0.005: 0.649 -> 0.677 | Improved |
| use auto-decoder instead of auto-encoder to regress TSDF-Def (ablation) | NeCGS · Mixed | RD curve | Improved |
| add SSIM term to regression loss (ablation) | NeCGS · Mixed | RD curve | Improved |
| raise TSDF-Def resolution from 32 to 64 (ablation) | NeCGS · AMA | CD (x1e-3): 16.982 -> 4.271 | Improved |
| raise TSDF-Def resolution from 128 to 256 (ablation) | NeCGS · AMA | CD (x1e-3): 3.436 -> 3.234 | Improved |
| raise TSDF-Def resolution from 128 to 256 (ablation) | NeCGS · AMA | inference time (ms): 98.95 -> 421.94 | Hurt |
| raise TSDF-Def resolution from 128 to 256 (ablation) | NeCGS · AMA | compression ratio: 253.45 -> 232.58 | Hurt |
| use NeCGS instead of V-PCC for set compression time | compression pipeline · Efficiency | compression time (h): 39.34 -> 10.01 | Improved |
| use NeCGS instead of Draco for compression time | compression pipeline · Efficiency | compression time (h): 0.06 -> 10.01 | Hurt |
Closest papers by meaning
- Neural Compression of 3D Meshes using Sparse Implicit Representation (ICLR 2026 · similarity 0.91)
- Cut Less, Fold More: Model Compression through the Lens of Projection Geometry (ICLR 2026 · similarity 0.87)
- ContextGS : Compact 3D Gaussian Splatting with Anchor Level Context Model (NeurIPS 2024 · similarity 0.86)
- CAT-3DGS: A Context-Adaptive Triplane Approach to Rate-Distortion-Optimized 3DGS Compression (ICLR 2025 · similarity 0.86)
- ZPressor: Bottleneck-Aware Compression for Scalable Feed-Forward 3DGS (NeurIPS 2025 · similarity 0.85)
- MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · 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.
