Research map / 3D vision / Face and shape
Rethinking Reprojection: Closing the Loop for Pose-Aware Shape Reconstruction From a Single Image
Rui Zhu, Hamed Kiani Galoogahi, Chaoyang Wang, Simon Lucey
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 |
|---|---|---|---|
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-TL · PASCAL3D+ aeroplane | 2D AP: 0.589 -> 0.704 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ aeroplane | 2D AP: 0.627 -> 0.720 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ chair | 2D AP: 0.852 -> 0.878 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ aeroplane | 3D AP: 0.183 -> 0.249 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-TL · PASCAL3D+ aeroplane | 3D AP: 0.211 -> 0.219 | No clear effect |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-TL · PASCAL3D+ aeroplane | median rotation error: 23.0 -> 17.3 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-TL · PASCAL3D+ car | rotation accuracy: 0.83 -> 0.80 | Hurt |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ car | translation median error: 0.061 -> 0.050 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-TL · PASCAL3D+ aligned shapes aeroplane | 3D AP: 0.552 -> 0.580 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ aligned shapes chair | 3D AP: 0.727 -> 0.763 | Improved |
| fine-tune on natural images with silhouette reprojection loss (ablation) | p-3D-VAE-GAN · PASCAL3D+ aligned shapes aeroplane | 3D AP: 0.669 -> 0.676 | No clear effect |
| decouple pose from style with separate pose regressor (p-3D-VAE-GAN vs 3D-VAE-GAN) | 3D-VAE-GAN · Rendered, pose-aware shapes im2V | AP: 0.403 -> 0.665 | Improved |
Closest papers by meaning
- Shape-Pose Ambiguity in Learning 3D Reconstruction from Images (AAAI 2021 · similarity 0.91)
- Neural Reprojection Error: Merging Feature Learning and Camera Pose Estimation (CVPR 2021 · similarity 0.90)
- SDFit: 3D Object Pose and Shape by Fitting a Morphable SDF to a Single Image (ICCV 2025 · similarity 0.89)
- Image-Based 3D Object Reconstruction: State-of-the-Art and Trends in the Deep Learning Era (TPAMI 2019 · similarity 0.88)
- Self-supervised Single-view 3D Reconstruction via Semantic Consistency (ECCV 2020 · similarity 0.88)
- GeoReF: Geometric Alignment Across Shape Variation for Category-level Object Pose Refinement (CVPR 2024 · similarity 0.87)
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.
