Research map / 3D vision / Human pose estimation
GP2C: Geometric Projection Parameter Consensus for Joint 3D Pose and Focal Length Estimation in the Wild
Alexander Grabner, Peter M. Roth, Vincent Lepetit
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 joint 3D pose and focal length refinement to the initial EPnP solution (Ours-LF) (ablation) | Ours-LF · Pix3D mean | Acc_P0.1: 84.7% -> 85.5% | Improved |
| add joint 3D pose and focal length refinement to the initial EPnP solution (Ours-BB) (ablation) | Ours-BB · Pix3D mean | Acc_P0.1: 86.8% -> 88.0% | Improved |
| add joint refinement (MedErr_P) (ablation) | Ours-BB · Pix3D mean | MedErr_P x10^2: 3.87 -> 3.66 | Improved |
| use Cauchy loss PnP instead of standard squared loss (ablation) | Ours-LF · Pix3D mean | Acc_P0.1: 85.3% -> 85.5% | No clear effect |
| use Cauchy loss PnP instead of standard squared loss (ablation) | Ours-BB · Pix3D mean | Acc_P0.1: 87.5% -> 88.0% | No clear effect |
| use RANSAC PnP instead of standard PnP (ablation) | Ours-BB · Pix3D mean | MedErr_P x10^2: 3.68 -> 3.68 | No clear effect |
| explicitly estimate and integrate focal length in pose estimation | Ours-LF vs prior SOTA · Pix3D mean | MedErr_t x10^1: 2.52 -> 1.85 | Improved |
| explicitly estimate and integrate focal length in pose estimation | Ours-BB vs prior SOTA · Pix3D mean | Acc_P0.1: 71.5% -> 88.0% | Improved |
| explicitly estimate and integrate focal length in pose estimation | Ours-BB vs prior SOTA · Comp car | MedErr_t x10^1: 3.30 -> 2.55 | Improved |
| explicitly estimate and integrate focal length in pose estimation | Ours-BB vs prior SOTA · Stanford car | Acc_P0.1: 76.4% -> 96.5% | Improved |
| initialize geometric optimization with predicted focal length instead of constant median focal length (ablation) | GP2C · Pix3D | Acc_R,t | Improved |
| initialize with ground truth focal length instead of predicted (ablation) | GP2C · Pix3D | Acc_R,t | Improved |
Closest papers by meaning
- Focal Length and Object Pose Estimation via Render and Compare (CVPR 2022 · similarity 0.90)
- FocalPose++: Focal Length and Object Pose Estimation via Render and Compare (TPAMI 2024 · similarity 0.89)
- 3D Human Pose Estimation = 2D Pose Estimation + Matching (CVPR 2017 · similarity 0.87)
- GIVEPose: Gradual Intra-class Variation Elimination for RGB-based Category-Level Object Pose Estimation (CVPR 2025 · similarity 0.86)
- LEAP: Liberate Sparse-View 3D Modeling from Camera Poses (ICLR 2024 · similarity 0.85)
- SchemaPose: RGB-Based Category-Level Pose Estimation with Parametric Category Schema (NeurIPS 2026 · 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.
