Research map / 3D vision / Human pose estimation
Exploring intermediate representation for monocular vehicle pose estimation
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 |
|---|---|---|---|
| Add intermediate geometrical representations instead of directly regressing pose angles (ablation) | Ego-Net pose estimation · KITTI validation (AP2D=100.00), Easy, Moderate, and Hard | AOS: 95.22 -> 99.58; 92.14 -> 99.06; 88.37 -> 96.55 | Helped |
| Use cross-ratio loss with unlabeled training instances (ablation) | Ego-Net coordinate localization · KITTI validation, PCK@0.1 | PCK@0.1: 31.3 -> 31.9 | Helped |
| Use cross-ratio loss with unlabeled training instances (ablation) | Ego-Net coordinate localization · KITTI validation, PCK@0.2 | PCK@0.2: 64.1 -> 64.8 | Helped |
| Use cross-ratio loss with unlabeled training instances (ablation) | Ego-Net coordinate localization · KITTI validation, PCK@0.3 | PCK@0.3: 79.3 -> 79.6 | Helped |
| Use cross-ratio loss with unlabeled training instances (ablation) | Ego-Net · KITTI validation, AOS across Easy, Moderate, and Hard | AOS: 99.75 -> 99.58; 98.95 -> 99.06; 96.48 -> 96.55 | Mixed |
| Replace the original pose-estimation branch with Ego-Net | M3D-RPN · KITTI validation, AOS Easy, Moderate, and Hard | AOS: 88.79 -> 90.20; 81.25 -> 83.60; 65.37 -> 67.53 | Helped |
| Replace the original pose-estimation branch with Ego-Net | D4LCN · KITTI validation, AOS Easy, Moderate, and Hard | AOS: 91.74 -> 92.62; 82.96 -> 84.25; 66.45 -> 67.60 | Helped |
| Add Ego-Net to the D4LCN system | D4LCN · KITTI validation, APBEV Easy, Moderate, and Hard | APBEV: 31.53 -> 33.60; 22.58 -> 25.38; 17.87 -> 22.80 | Helped |
Closest papers by meaning
- Mono3D++: Monocular 3D Vehicle Detection with Two-Scale 3D Hypotheses and Task Priors (AAAI 2019 · similarity 0.88)
- Geometry-Guided Domain Generalization for Monocular 3D Object Detection (AAAI 2024 · similarity 0.85)
- UNOPose: Unseen Object Pose Estimation with an Unposed RGB-D Reference Image (CVPR 2025 · similarity 0.85)
- Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis (NeurIPS 2026 · similarity 0.84)
- Co-op: Correspondence-based Novel Object Pose Estimation (CVPR 2025 · similarity 0.84)
- GIVEPose: Gradual Intra-class Variation Elimination for RGB-based Category-Level Object Pose Estimation (CVPR 2025 · similarity 0.84)
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.