Research map / Detection and segmentation / 3D detection and driving
Recognizing Objects From Any View With Object and Viewer-Centered Representations
Sainan Liu, Vincent Nguyen, Isaac Rehg, Zhuowen Tu
Check 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 |
|---|---|---|---|
| Combine object-centered and viewer-centered branches | OVCNet · gMIRO SeenInstances | accuracy overall (%): 68.34 -> 73.24 | Improved |
| Combine object-centered and viewer-centered branches | OVCNet · gMIRO NovelInstances | accuracy overall (%): 53.27 -> 65.85 | Improved |
| Add object-centered, VC (3D), and VC (2D) branches together (ablation) | OVCNet · gMIVO SeenInstances | accuracy (%): 63.66 -> 79.36 | Improved |
| Add object-centered, VC (3D), and VC (2D) branches together (ablation) | OVCNet · gMIVO NovelInstances | accuracy (%): 64.50 -> 75.33 | Improved |
| Add the ancillary spherical pathway integrating VC (3D) features (ablation) | OC module · gMIRO SeenInstances | accuracy overall (%): 44.62 ± 0.58 -> 54.62 ± 0.73 | Improved |
| Add texture spherical-map information (ablation) | GenRe + spherical CNNs (bw=112) · gMIRO SeenInstances | accuracy overall (%): 42.43 ± 1.24 -> 44.62 ± 0.58 | Improved |
| Increase spherical CNN bandwidth from 60 to 112 (ablation) | GenRe + spherical CNNs · gMIRO SeenInstances | accuracy overall (%): 40.79 ± 1.21 -> 42.43 ± 1.24 | Improved |
| Increase spherical CNN bandwidth from 112 to 128 (ablation) | GenRe + spherical CNNs · gMIRO SeenInstances | accuracy overall (%): 42.43 ± 1.24 -> 40.94 ± 1.88 | Hurt |
| Increase voxel resolution from 30×30×30 to 128×128×128 (ablation) | GenRe + 3D CNNs · gMIRO SeenInstances | accuracy overall (%): 20.94 ± 0.41 -> 30.26 ± 0.62 | Improved |
| Increase 3D viewpoint augmentations from 1 to 160 (ablation) | GenRe + ResNet18 · gMIRO SeenInstances | accuracy overall (%): 32.49 ± 0.68 -> 45.15 ± 0.46 | Improved |
| Increase 3D viewpoint augmentations from 160 to 640 (ablation) | GenRe + ResNet18 · gMIRO SeenInstances | accuracy overall (%): 45.15 ± 0.46 -> 51.24 ± 0.23 | Improved |
| Use 30-degree rather than denser in-plane rotation augmentation (ablation) | VC (2D) ResNet18 · gMIRO | accuracy | No clear effect |
Closest papers by meaning
- Beyond Viewpoint: Robust 3D Object Recognition under Arbitrary Views through Joint Multi-Part Representation (ECCV 2024 · similarity 0.87)
- Learning Relationships for Multi-View 3D Object Recognition (ICCV 2019 · similarity 0.87)
- Pixels, Voxels, and Views: A Study of Shape Representations for Single View 3D Object Shape Prediction (CVPR 2018 · similarity 0.87)
- RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews From Unsupervised Viewpoints (CVPR 2018 · similarity 0.87)
- RotationNet for Joint Object Categorization and Unsupervised Pose Estimation from Multi-View Images (TPAMI 2019 · similarity 0.86)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · 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.
