Research map / 3D vision / Face and shape
Deep Optimized Priors for 3D Shape Modeling and Reconstruction
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 |
|---|---|---|---|
| Jointly optimize the pre-trained prior parameters and latent code instead of fixing the prior | Deep Optimized Priors · Shape auto-encoding, Chair and Table | mean and median CD (10^-3): mean: 0.21 -> 0.08 (Chair), 0.42 -> 0.10 (Table); median: 0.08 -> 0.06 (Chair), 0.07 -> 0.05 (Table) | Helped |
| Jointly optimize the pre-trained prior parameters and latent code according to image constraints | Deep Optimized Priors · Sparse multi-view reconstruction, Chair and Lamp | mean CD (10^-3): Chair: 2.75 -> 1.75 (DISN); Lamp: 6.32 -> 5.44 (LSM) | Helped |
| Jointly optimize the pre-trained prior parameters and latent code according to image constraints | Deep Optimized Priors · Sparse multi-view reconstruction, Chair and Lamp | F-score: Chair: 52.47 -> 62.34 (DISN); Lamp: 32.15 -> 36.20 (P2M++) | Helped |
| Optimize the pre-trained prior parameters and latent code to fit sparse point clouds | Deep Optimized Priors · Sparse point cloud reconstruction, Chair and Car | mean CD (10^-3): Chair: 0.72 -> 0.64 (IFNet); Car: 0.72 -> 0.28 (IGR) | Helped |
| Optimize the pre-trained prior parameters and latent code to fit sparse point clouds | Deep Optimized Priors · Sparse point cloud reconstruction, Chair and Car | Normal consistency: Chair: 0.84 -> 0.86 (IFNet); Car: 0.89 -> 0.90 (IGR) | Helped |
| Optimize only the latent code and keep the pre-trained network parameters fixed (ablation) | Deep Optimized Priors · Single-view multi-view reconstruction | reconstruction quality | Hurt |
| Replace the pre-trained network initialization with randomly initialized parameters (ablation) | Deep Optimized Priors · Single-view multi-view reconstruction | reconstruction plausibility | Hurt |
| Remove the L2 constraint on network parameters (ablation) | Deep Optimized Priors · Sparse multi-view reconstruction | general-view reconstruction quality | Hurt |
| Increase the number of input views (ablation) | Deep Optimized Priors · Multi-view reconstruction | reconstruction quality | Helped |
Closest papers by meaning
- Fostering Generalization in Single-View 3D Reconstruction by Learning a Hierarchy of Local and Global Shape Priors (CVPR 2021 · similarity 0.92)
- AutoSDF: Shape Priors for 3D Completion, Reconstruction and Generation (CVPR 2022 · similarity 0.90)
- Flow-Corrected Shape Optimization: Taming Manifold Drift in High-Dimensional 3D Models (NeurIPS 2026 · similarity 0.88)
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 2025 · similarity 0.88)
- GPLD3D: Latent Diffusion of 3D Shape Generative Models by Enforcing Geometric and Physical Priors (CVPR 2024 · similarity 0.88)
- SPAR3D: Stable Point-Aware Reconstruction of 3D Objects from Single Images (CVPR 2025 · similarity 0.88)
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.