Research map / 3D vision / Face and shape
DPF-Net: Combining Explicit Shape Priors in Deformable Primitive Field for Unsupervised Structural Reconstruction of 3D Objects
Qingyao Shuai, Chi Zhang, Kaizhi Yang, Xuejin Chen
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 primitive deformation module on top of parameterized primitives (ablation) | DPF-Net · ShapeNet table | Chamfer Distance (x0.001): 0.7356 -> 0.4756 | Improved |
| add primitive deformation module (ablation) | DPF-Net · ShapeNet airplane | m-IoU: 56.3 -> 66.0 | Improved |
| add deformation on top of intermediate primitives only (PPM) for reconstruction (ablation) | DPF-Net · ShapeNet chair | Chamfer Distance (x0.001): 0.8127 -> 0.3633 | Improved |
| use cylinder instead of cuboid primitives (ablation) | DPF-Net · ShapeNet table | Chamfer Distance / m-IoU: CD 0.4756 -> 0.4251; m-IoU 91.3 -> 90.4 | Mixed |
| use DPF-Net instead of BSP-Net on airplane segmentation | DPF-Net · ShapeNet airplane segmentation | m-IoU: 74.2 -> 66.0 | Hurt |
| DPF-Net vs BSP-Net in reconstruction CD | DPF-Net · ShapeNet mean | Chamfer Distance (x0.001): 0.3303 -> 0.3438 | No clear effect |
| use DPF-Net for single-view reconstruction | DPF-Net · ShapeNet single-view mean | Chamfer Distance (x0.001): 0.7201 -> 0.6847 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- RIM-Net — RIM-Net: Recursive Implicit Fields for Unsupervised Learning of Hierarchical Shape Structures (CVPR 2022) · better on 5 of 5 shared results
- BSP-Net — BSP-Net: Generating Compact Meshes via Binary Space Partitioning (CVPR 2020) · better on 2 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
- Residual Primitive Fitting of 3D Shapes with SuperFrusta (CVPR 2026 · similarity 0.86)
- PyTorchGeoNodes: Enabling Differentiable Shape Programs for 3D Shape Reconstruction (CVPR 2025 · similarity 0.86)
- D2IM-Net: Learning Detail Disentangled Implicit Fields From Single Images (CVPR 2021 · similarity 0.86)
- RIM-Net: Recursive Implicit Fields for Unsupervised Learning of Hierarchical Shape Structures (CVPR 2022 · similarity 0.86)
- Unified Primitive Proxies for Structured Shape Completion (CVPR 2026 · similarity 0.86)
- Unsupervised 3D Shape Reconstruction by Part Retrieval and Assembly (CVPR 2023 · 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.
