Research map / Point clouds / Point cloud classification
PointNetVLAD: Deep Point Cloud Based Retrieval for Large-Scale Place Recognition
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 |
|---|---|---|---|
| Replace PointNet max pooling with NetVLAD aggregation | PointNet place-recognition descriptor · Oxford | average recall @ top 1%: 73.44 -> 80.31 | Helped |
| Replace PointNet max pooling with NetVLAD aggregation | PointNet place-recognition descriptor · U.S. | average recall @ top 1%: 64.64 -> 72.63 | Helped |
| Replace PointNet max pooling with NetVLAD aggregation | PointNet place-recognition descriptor · R.A. | average recall @ top 1%: 51.92 -> 60.27 | Helped |
| Replace PointNet max pooling with NetVLAD aggregation | PointNet place-recognition descriptor · B.D. | average recall @ top 1%: 54.74 -> 65.30 | Helped |
| Use lazy triplet loss rather than standard triplet loss (ablation) | PN_VLAD · Oxford | average recall @ top 1%: 71.20 -> 78.99 | Helped |
| Use lazy quadruplet loss rather than standard quadruplet loss (ablation) | PN_VLAD · Oxford | average recall @ top 1%: 74.13 -> 80.31 | Helped |
| Add the second, randomly sampled negative-separation term of lazy quadruplet loss (ablation) | PN_VLAD with lazy triplet loss · Oxford | average recall @ top 1%: 78.99 -> 80.31 | Helped |
| Increase descriptor dimension from 128 to 256 (ablation) | PN_VLAD · Oxford | average recall @ top 1%: 74.60 -> 80.31 | Helped |
| Increase descriptor dimension from 256 to 512 (ablation) | PN_VLAD · Oxford | average recall @ top 1%: 80.31 -> 80.33 | No clear effect |
Closest papers by meaning
- SOE-Net: A Self-Attention and Orientation Encoding Network for Point Cloud Based Place Recognition (CVPR 2021 · similarity 0.91)
- PCAN: 3D Attention Map Learning Using Contextual Information for Point Cloud Based Retrieval (CVPR 2019 · similarity 0.89)
- NetVLAD: CNN Architecture for Weakly Supervised Place Recognition (CVPR 2016 · similarity 0.89)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.87)
- PointRWKV: Efficient RWKV-Like Model for Hierarchical Point Cloud Learning (AAAI 2025 · similarity 0.87)
- DeepLA-Net: Very Deep Local Aggregation Networks for Point Cloud Analysis (CVPR 2025 · 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.