Research map / Point clouds / Segmentation and detection
Deep FusionNet for Point Cloud Semantic Segmentation
Feihu Zhang Jin Fang Benjamin Wah Philip Torr
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 neighborhood-voxel point feature aggregation to voxel convolution (ablation) | sparse conv baseline · Semantic KITTI | mean IoU (%): 54.8 -> 58.9 | Improved |
| add inner-voxel aggregation (ablation) | FusionNet without inner-voxel · Semantic KITTI | mean IoU (%): 58.9 -> 63.7 | Improved |
| add inner-voxel aggregation (ablation) | FusionNet without inner-voxel · Semantic KITTI | mean accuracy (%): 90.7 -> 91.8 | Improved |
| add voxel-level aggregation to point-only network (ablation) | point feature aggregation only · Semantic KITTI | mean IoU (%): 41.6 -> 63.7 | Improved |
| add point-level features to voxel-only network (full FusionNet) (ablation) | sparse conv baseline · Semantic KITTI | mean accuracy (%): 89.6 -> 91.8 | Improved |
| use FusionNet instead of sparse convolution MinkNet42 | MinkNet42 · Semantic KITTI test | mIoU (%): 54.3 -> 61.3 | Improved |
| use FusionNet instead of MinkNet | MinkNet · ScanNet | mIoU (%): 67.9 -> 68.8 | Improved |
| use FusionNet instead of KPConv | KPConv · S3DIS Area 5 | mIoU (%): 67.1 -> 67.2 | No clear effect |
| use FusionNet at 5cm voxel instead of sparse conv at 3cm voxel | sparse convolution · Semantic KITTI val | mean IoU | Improved |
| use finer voxel size (ablation) | voxel-based networks · Semantic KITTI val | mean IoU / memory | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- PointConv — PointConv: Deep Convolutional Networks on 3D Point Clouds (CVPR 2019) · better on 3 of 3 shared results
- PointCNN — PointCNN: Convolution On X-Transformed Points (NeurIPS 2018) · better on 2 of 2 shared results
- PointASNL — PointASNL: Robust Point Clouds Processing Using Nonlocal Neural Networks With Adaptive Sampling (CVPR 2020) · better on 1 of 1 shared result
- RandLANet — RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (CVPR 2020) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (CVPR 2017 · similarity 0.91)
- Learning Semantic Segmentation of Large-Scale Point Clouds with Random Sampling (TPAMI 2021 · similarity 0.90)
- Fully-Convolutional Point Networks for Large-Scale Point Clouds (ECCV 2018 · similarity 0.90)
- RPVNet: A Deep and Efficient Range-Point-Voxel Fusion Network for LiDAR Point Cloud Segmentation (ICCV 2021 · similarity 0.90)
- Pyramid Architecture for Multi-Scale Processing in Point Cloud Segmentation (CVPR 2022 · similarity 0.90)
- RandLA-Net: Efficient Semantic Segmentation of Large-Scale Point Clouds (CVPR 2020 · similarity 0.90)
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.
