Research map / Point clouds / Point cloud classification
End-to-End Learning Local Multi-View Descriptors for 3D Point Clouds
Lei Li, Siyu Zhu, Hongbo Fu, Ping Tan, Chiew-Lan Tai
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 |
|---|---|---|---|
| Use end-to-end optimizable viewpoints instead of random sampling (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 64.8 -> 79.9 | Improved |
| Use end-to-end optimizable viewpoints instead of viewpoint clustering (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 53.3 -> 79.9 | Improved |
| Use end-to-end optimizable viewpoints instead of orbited placement (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 42.2 -> 79.9 | Improved |
| Add in-plane rotation augmentation to rendered view patches (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 54.9 -> 79.9 | Improved |
| Replace max-view pooling with soft-view pooling (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 66.8 -> 79.9 | Improved |
| Replace Fuseption with soft-view pooling (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 55.9 -> 79.9 | Improved |
| Replace NetVLAD with soft-view pooling (ablation) | Local multi-view descriptors · ETH | average recall (τ₂=0.05, %): 58.7 -> 79.9 | Improved |
| Replace max-view pooling with soft-view pooling (ablation) | Local multi-view descriptors · 3DMatch | average recall (τ₂=0.05, %): 96.9 -> 97.5 | Improved |
| Increase descriptor dimension to d ≥ 32 (ablation) | Local multi-view descriptors · 3DMatch | average recall | No clear effect |
| Increase viewpoint number to n ≥ 8 (ablation) | Local multi-view descriptors · 3DMatch | average recall | No clear effect |
Closest papers by meaning
- Learning and Matching Multi-View Descriptors for Registration of Point Clouds (ECCV 2018 · similarity 0.90)
- Learning general and distinctive 3D local deep descriptors for point cloud registration (TPAMI 2022 · similarity 0.90)
- Unsupervised Learning of Local Equivariant Descriptors for Point Clouds (TPAMI 2021 · similarity 0.89)
- DH3D: Deep Hierarchical 3D Descriptors for Robust Large-Scale 6DoF Relocalization (ECCV 2020 · similarity 0.89)
- ViewNet: A Novel Projection-Based Backbone With View Pooling for Few-Shot Point Cloud Classification (CVPR 2023 · similarity 0.88)
- Multi-View Representation is What You Need for Point-Cloud Pre-Training (ICLR 2024 · similarity 0.88)
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.
