Research map / 3D vision / Face and shape
Learning 3D Object Categories by Looking Around Them
David Novotny, Diane Larlus, Andrea Vedaldi
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 probabilistic losses and confidence prediction (ablation) | VpDR-Net · chair, LDOS | angular error eR: 45.33 -> 33.92 | Improved |
| add probabilistic losses and confidence prediction (ablation) | VpDR-Net · chair, Pascal3D | angular error eR: 57.23 -> 42.34 | Improved |
| add depth and point cloud prediction branches during training (ablation) | VpDR-Net · chair, LDOS | angular error eR: 68.19 -> 33.92 | Improved |
| add depth and point cloud prediction branches during training (ablation) | VpDR-Net · chair, Pascal3D | angular error eR: 60.31 -> 42.34 | Improved |
| add geometry-aware data augmentation (ablation) | VpDR-Net · chair, LDOS | angular error eR: 35.16 -> 33.92 | Improved |
| add geometry-aware data augmentation (ablation) | VpDR-Net · chair, Pascal3D | angular error eR: 43.52 -> 42.34 | Improved |
| add geometry-aware data augmentation (ablation) | VpDR-Net · chair, Pascal3D | APeR: 0.10 -> 0.23 | Improved |
| use viewpoint factorization training instead of aligning shapes in 3D | VpDR-Net vs VPNet-unsupervised · car, Pascal3D | angular error eR: 49.62 -> 29.57 | Improved |
| use viewpoint factorization training instead of aligning shapes in 3D | VpDR-Net vs VPNet-unsupervised · chair, LDOS | angular error eR: 30.56 -> 33.92 | Hurt |
| use VpDR-Net confidence instead of VPNet softmax confidence | VpDR-Net · car, Pascal3D | APeC: 0.01 -> 0.91 | Improved |
| predict point clouds with VpDR-Net instead of retrieval (Aubry) | VpDR-Net · LDOS chairs point cloud | mDpcl: 1.30 -> 0.20 | Improved |
| fuse predicted point cloud with partial depth point cloud | VpDR-Net-Fuse · FrC cars point cloud | mDpcl: 0.28 -> 0.26 | Improved |
Closest papers by meaning
- Capturing the Geometry of Object Categories from Video Supervision (TPAMI 2018 · similarity 0.91)
- Planes vs. Chairs: Category-Guided 3D Shape Learning without Any 3D Cues (ECCV 2022 · similarity 0.88)
- Learning 3D Object Shape and Layout Without 3D Supervision (CVPR 2022 · similarity 0.88)
- Learning to Group: A Bottom-Up Framework for 3D Part Discovery in Unseen Categories (ICLR 2020 · similarity 0.87)
- Weakly-Supervised Learning of Category-Specific 3D Object Shapes (TPAMI 2019 · similarity 0.87)
- Unsupervised Learning of Category-Level 3D Pose from Object-Centric Videos (CVPR 2024 · 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.
