Research map / Point clouds / Point cloud classification
CP3: Channel Pruning Plug-In for Point-Based Networks
Yaomin Huang, Ning Liu, Zhengping Che, Zhiyuan Xu, Chaomin Shen, Yaxin Peng, Guixu Zhang, Xinmei Liu, Feifei Feng, Jian Tang
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 CP³ to HRank channel pruning | PointNet++ · ModelNet40 test | OA: 92.59 -> 92.95 | Improved |
| Add CP³ to ResRep channel pruning | PointNeXt-S (C=64) · ScanObjectNN test | OA: 86.58 -> 88.52 | Improved |
| Add CP³ to CHIP channel pruning | PointNeXt-B · S3DIS Area-5 | mIoU: 66.57 -> 66.80 | Improved |
| Add CP³ to ResRep channel pruning | VoteNet · ScanNetV2 | mAP@0.25: 62.45 -> 63.92 | Improved |
| Add CP³ to ResRep channel pruning | VoteNet · SUN RGB-D | mAP@0.50: 36.80 -> 37.37 | Improved |
| Add CP³ to ResRep channel pruning | GroupFree3D · ScanNetV2 | mAP@0.25: 67.21 -> 68.57 | Improved |
| Add knowledge recycling (KR) without CE (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 75% | OA: 84.79 -> 85.63 | Improved |
| Add coordinate enhancement (CE) without KR (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 75% | OA: 84.79 -> 85.11 | Improved |
| Add CE and KR together (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 75% | OA: 84.79 -> 86.63 | Improved |
| Add knowledge recycling (KR) without CE (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 90% | OA: 81.33 -> 83.66 | Improved |
| Add coordinate enhancement (CE) without KR (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 90% | OA: 81.33 -> 83.10 | Improved |
| Add CE and KR together (ablation) | HRank-pruned PointNeXt-S (C=32) · ScanObjectNN test, FLOPs drop 90% | OA: 81.33 -> 84.83 | Improved |
Closest papers by meaning
- Accelerate CNNs from Three Dimensions: A Comprehensive Pruning Framework (ICML 2021 · similarity 0.89)
- Revisiting Random Channel Pruning for Neural Network Compression (CVPR 2022 · similarity 0.87)
- Efficient Distortion-Minimized Layerwise Pruning (TPAMI 2025 · similarity 0.86)
- REPrune: Channel Pruning via Kernel Representative Selection (AAAI 2024 · similarity 0.86)
- BilevelPruning: Unified Dynamic and Static Channel Pruning for Convolutional Neural Networks (CVPR 2024 · similarity 0.86)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (AAAI 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.
