Research map / 3D vision / Human pose estimation
RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews From Unsupervised Viewpoints
Read the paperCheck 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 estimate category and latent viewpoints using RotationNet | AlexNet on ModelNet40 · ModelNet40 | classification accuracy: 93.8 -> 97.37 | Helped |
| jointly estimate category and latent viewpoints using RotationNet | AlexNet on ModelNet10 · ModelNet10 | classification accuracy: 94.0 -> 98.46 | Helped |
| use 12 views instead of a single view (ablation) | RotationNet on RGBD object benchmark · RGBD object benchmark | instance recognition accuracy: 90.44 -> 97.45 | Helped |
| use 12 views instead of a single view (ablation) | RotationNet on RGBD object benchmark · RGBD object benchmark | category recognition accuracy: 96.55 -> 99.51 | Helped |
| use 12 views instead of a single view (ablation) | RotationNet on RGBD object benchmark · RGBD object benchmark | average pose accuracy: 78.67 -> 81.17 | Helped |
| replace VGG-M with AlexNet (ablation) | RotationNet on ModelNet40 · ModelNet40 | classification accuracy: 94.68 ± 1.16 -> 93.70 ± 1.07 | Hurt |
| replace VGG-M with ResNet-50 (ablation) | RotationNet on ModelNet40 · ModelNet40 | classification accuracy: 94.68 ± 1.16 -> 94.77 ± 1.10 | No clear effect |
| replace VGG-M with AlexNet (ablation) | RotationNet on ModelNet10 · ModelNet10 | classification accuracy: 94.82 ± 1.17 -> 94.52 ± 1.01 | No clear effect |
| replace VGG-M with ResNet-50 (ablation) | RotationNet on ModelNet10 · ModelNet10 | classification accuracy: 94.82 ± 1.17 -> 94.80 ± 0.96 | No clear effect |
| test randomly rotated models after training on aligned models (ablation) | RotationNet on ShapeNetCore55 · ShapeNetCore55 | classification accuracy: 90% -> 38% | Hurt |
| train with unaligned models rather than aligned models (ablation) | RotationNet on ShapeNetCore55 · ShapeNetCore55 unaligned test models | classification accuracy: 38% -> 73% | Helped |
| fine-tune ImageNet-pretrained weights instead of training without fine-tuning (ablation) | RotationNet (AlexNet) on ModelNet40 · ModelNet40 | classification accuracy | Helped |
Closest papers by meaning
- A Comparative Analysis and Study of Multiview CNN Models for Joint Object Categorization and Pose Estimation (ICML 2016 · similarity 0.92)
- Multi-Path Learning for Object Pose Estimation Across Domains (CVPR 2020 · similarity 0.89)
- Unsupervised Learning of Category-Level 3D Pose from Object-Centric Videos (CVPR 2024 · similarity 0.88)
- NOPE: Novel Object Pose Estimation from a Single Image (CVPR 2024 · similarity 0.88)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.88)
- LEAP: Liberate Sparse-View 3D Modeling from Camera Poses (ICLR 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.