Research map / 3D vision / Human pose estimation
Category Level Object Pose Estimation via Neural Analysis-by-Synthesis
Xu Chen, Zijian Dong, Jie Song, Andreas Geiger, Otmar Hilliges
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 3D feature volume to the generator (ablation) | pose-aware generator · ShapeNet pose fitting | AP60: 90.2% -> 97.1% | Improved |
| add 3D feature volume to the generator (ablation) | pose-aware generator · ShapeNet pose fitting | AP10: 1.8% -> 16.5% | Improved |
| add KL divergence (VAE) training (ablation) | pose-aware generator · ShapeNet pose fitting | AP10: 12.3% -> 16.5% | Improved |
| add KL divergence (VAE) training (ablation) | pose-aware generator · ShapeNet pose fitting | L1 reconstruction error: 0.049 -> 0.056 | Hurt |
| use 16-dim latent space instead of 4 (ablation) | pose-aware generator · ShapeNet pose fitting | AP10: 9.1% -> 16.5% | Improved |
| use 16-dim latent space instead of 128 (ablation) | pose-aware generator · ShapeNet pose fitting | AP10: 11.4% -> 16.5% | Improved |
| increase latent dimension from 16 to 128 (ablation) | pose-aware generator · ShapeNet reconstruction | L1 reconstruction error: 0.056 -> 0.041 | Improved |
| use perceptual loss instead of L1 energy (ablation) | pose fitting · ShapeNet pose fitting | AP60: 85.3% -> 97.1% | Improved |
| use perceptual loss instead of L2 energy (ablation) | pose fitting · ShapeNet pose fitting | AP60: 88.3% -> 97.1% | Improved |
| use perceptual loss instead of SSIM energy (ablation) | pose fitting · ShapeNet pose fitting | AP60: 92.5% -> 97.1% | Improved |
| add latent-code regularization to the energy (ablation) | pose fitting · ShapeNet pose fitting | AP60: 95.9% -> 97.1% | Improved |
Closest papers by meaning
- Rethinking Correspondence-based Category-Level Object Pose Estimation (CVPR 2025 · similarity 0.88)
- SchemaPose: RGB-Based Category-Level Pose Estimation with Parametric Category Schema (NeurIPS 2026 · similarity 0.87)
- Co-op: Correspondence-based Novel Object Pose Estimation (CVPR 2025 · similarity 0.87)
- NOPE: Novel Object Pose Estimation from a Single Image (CVPR 2024 · similarity 0.86)
- GeoReF: Geometric Alignment Across Shape Variation for Category-level Object Pose Refinement (CVPR 2024 · similarity 0.86)
- GIVEPose: Gradual Intra-class Variation Elimination for RGB-based Category-Level Object Pose Estimation (CVPR 2025 · similarity 0.86)
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.
